TEAM · Thesis Check · 2026.10.20
Seats 从"下滑"重置为"增长"，证据站得住吗
用 Atlassian 18 场管理层发言和 Knowledge Base 里的专家、卖方材料，逐条核对 Caffrey 的 seats 增长 thesis。原文在前，分析在后。
54份材料
18场管理层发言
462条逐字原文
0个 seat 数字
总体判断
口径：用 Atlassian 自己的 17 场业绩会和卖方会议 (2025.05–2026.09)、1 份 FactSet 会议转录，以及 Knowledge Base 里的专家访谈和卖方摘要，逐条核对"TEAM 的 seats 从预期下滑重置为持续增长"这个 thesis。每条引文都已用脚本对原文逐字核对，并补成完整原文段落 (不删节)，带文件名和行号。
Executive Summary
Seats 确实还在涨，这一点管理层每个季度都说，渠道商和前员工也没看到减少；但公司从来没给过 seat 数或增速，CFO 的原话是 seat 扩张速率"稳定了 4-6 个季度"，是稳定，不是加速。
过去一年的加速主要来自三件可以拆开的事：上云迁移 (每年给云收入贡献中到高个位数的增长)、Teamwork Collection 打包和版本升级、每年 7-10% 的提价。"Coding agent 让更多人进来、所以 seats 加速"这条链，唯一的数字是"用 AI 编码工具的客户 seat 扩张快 5%"，CEO 自己承认这批是增长型的前沿公司。
AI 带来的工作量增长远快于 seat：经 MCP 建的 Jira 工作项是 4 倍，MCP 调用一个季度涨 400%，而人通过网页和手机用产品"几乎没变"。这部分价值要靠 credits 和用量去收，12 月 3 日才开始收，还没进数字。
FY27 subscription ARR 指引 18%，隐含有机净新增 ARR 大约持平 ($1.18B → $1.19B)；股价从 4 月低点已经涨了约 3.8 倍。"seats 不会消失"这个预期修复已经发生并且定价了；要再翻一倍，需要 seats 真正加速或用量收入起来，目前两者都没有证据。
逐条结论
| Thesis 环节 | 结论 | 证据强度 | 一句话 |
|---|---|---|---|
| E01 付费 seats 仍在净增长 | 成立 | 中 | 18 场全部说在涨，零个数字；"稳定"而非加速 |
| E02 增长来自非开发者 | 方向成立，数字不可比 | 弱-中 | 非开发者占比一年里从 48% 说到 65%，主要是口径变了 |
| E03 Coding agent 对 seats 是净利好 | 只证明"没变坏" | 弱 | 唯一数字"快 5%"有选择偏差；工作量涨得比 seat 快得多 |
| E04 打包和提价提升 ARPU | 成立 | 中-强 | TWC 100 万 seats、10 倍 credits、版本升级 +40%；但 TWC 升级时 seat 数不变 |
| E05 合成可持续的 secular 加速 | 证据不足 | 弱 | 迁移贡献持续到 FY29 以后，FY27 收入见底，20% CAGR 目标已撤 |
| E06 预期 | 预期修复已定价 | 强 | FY27 有机净新增 ARR 持平；股价 $57 → $215 |
对照你的两条标准
- 好公司：生意本身稳。seat 没有萎缩，Fortune 500 覆盖 85% 但只占约 10% 收入 (还有渗透空间)，Jira 每人每月 $7-8 便宜到很难被替换。问题在股东回报：SBC 和 FCF 差不多大 (我们的 primer 和 mapping 结论)，seat 增长再稳，落到股东手里的现金也不多
- 两年一倍：现价约 $215，翻倍是 $430。primer 的 bull case 是 $352 (base $166)。要到 $430，需要 seats 从"稳定"变成"加速"，或者用量收入成规模；这次核对的 54 份材料里两者都没有证据
- 按你的规则：现价两条都不满足，应该放弃，或者放进观察名单等价格。primer 的买点 $150-165 仍然适用
Follow-up
| 时间 | 盯什么 | 什么结果说明什么 |
|---|---|---|
| 2026-11-05 Q1 FY27 | 云收入 vs 指引 28.5%；管理层说 seat 扩张是"稳定"还是"加速"；有没有第一次给 seat 数 | 还是只说"strong"而不给数，说明拿不出加速的数字；净新增 ARR 跑赢"持平"的指引，才说明有机增长在加速 |
| 2026-12-03 起 | Rovo credit 超额收费开始 (Mizuho 估约 10% 云客户会被收费) | Q2/Q3 FY27 如果管理层开始单独报 credit 收入，说明 agent 工作量开始变现；只报 MAU 说明量很小 |
| 2027-02 Q2 FY27 | TWC 客户数和 seats 更新 (上次是 2026 年 2 月的 1,000 家 / 100 万) | TWC 增速放缓，说明打包带来的一次性提价在消化完 |
| 持续 | Agent 怎么收费：会不会给 agent 发 license，flex 承诺制推广到多大 | 如果走 flex 和 credits，seat 数会越来越不是收入的口径，模型要改按用量看 |
| 可以约的专家 | 渠道商或大客户 IT 采购：续约时 Jira/Confluence seat 同比变化；新增 seat 里非开发者占多少；有没有用 service account 接 agent | 这是目前所有材料里最缺的一手数据 |
证据限制：管理层材料占了证据的大头，而管理层每次说 seat 都不给数；专家里真正讲 Atlassian 的只有前 VP 和一家渠道商，前 VP 的核心论据来自他在 ServiceNow 的经历，渠道商的增长集中在国防单一客户；专家纪要大多只检索了相关段落，没有通读。
反常识 Check List
按对 thesis 的冲击大小排序。每条先说人话，再放原文。
1. FY27 指引隐含有机净新增 ARR 持平 → 冲击 E05/E06
23% 的 ARR 增速里约 1 个点来自收购 DX，指引 18% 意味着明年新增的 ARR 金额和今年差不多。Morgan Stanley 直接问是不是 seat 扩张在放缓，CFO 只讲审慎和宏观，没有回答。
Overnight success, 3 years in the making. I guess maybe just on the initial ARR guide, James, I think it implies roughly flat organic net new ARR in fiscal '27. And you said you're reflecting the same level of prudence that you have historic -- last historically in this outlook. Is there anything different exiting fiscal '26 that increases the confidence in those numbers given some of those dynamics that you guys just went over? Or just maybe just help level set that a little bit?
Thanks, Adam, for the question, and welcome to the coverage. First, I would highlight that we're really pleased to see the strength that we saw in the quarter, printing ARR growth, sub-ARR growth of 23% year-over-year. When it comes to the guide, we'll continue to take a thoughtful and prudent approach here, really balancing the momentum that we're seeing in the business, the demand signals that we're getting from our customers with the uncertainty as you look out in the macro environment, fiscal policy and geopolitical landscape. And as you mentioned, this is the first time we'll be issuing ARR guidance for the full fiscal year. We're also still early in the fiscal year here. So taking a bit more of a prudent approach in that second half of FY '27. It's also worth noting that in FY '26, the ARR year-over-year growth rate benefited by approximately 1 point from the acquisition of DX at the close of November in 2025. That's our fiscal Q2. And so we'll be lapping that DX acquisition and the healthy acceleration that we saw in the second half of FY '26. And of course, we're going to still continue to scale our enterprise go-to-market sales motion as we unlock the opportunities that we see in front of us. So all I feel really good about the state of the business here. And coming off a quarter where RPO landed at 44% year-on-year NRR was north of 120%. So you're just taking a balanced and prudent approach to the guide.
2. CFO：seat 扩张速率"稳定 4-6 个季度" → 冲击 E01 的"加速"
这是管理层对 seat 速度最具体的一句话。稳定说明没变坏，但也说明云收入从 20% 出头加速到 31% 不是 seat 加速带来的。
Yes. Thanks for the question, Ittai. In terms of where we see the expansion, I'd say it's broad-based. It's across both tech and non-tech users. We are making a lot of progress on what we call non-tech or business users, particularly with the TWC product. Those seat expansion rates, both across the enterprise and SMB remain stable. That's been the case for 4 to 6 quarters now. So we feel really good about the continued progress on that. And that's the way I would describe from a paid seat expansion perspective, sort of the color in terms of what we're seeing on that front.
Yes. Great question. Thanks for the question. So overall, we had great execution from our sales team and partners, which drove a very strong underlying bookings and billings results this quarter, as you pointed out. Mike highlighted a lot of these things in his opening remarks from the record number of deals with over $1 million to Data Center to Cloud migrations being up 60% year-over-year to the strong momentum we had in Teamwork Collection. All of that culminated in the RPO balance increasing to $3.3 billion. That's up 38% year-over-year. Roughly 74% of that balance will be recognized in revenue in the next 12 months, and that's up 29% year-over- year. And the remaining portion is increased 71%, and that's on the strength of large cloud multiyear agreements signed in the quarter, which is a great sign of customer commitment and confidence in our platform and our road map. So overall, a very strong quarter of progress and execution in the enterprise portion of our business, which drove the bookings and billings performance you see in the results. Now if you go down to revenue, that was also better than our expectations. As you saw in the numbers, we had outperformance on Cloud and marketplace and in-line performance on Data Center. On Cloud revenue guidance coming into the quarter, we'd assume some negative impact on paid seat expansion rates and cloud migrations from macro uncertainty and our Enterprise go-to-market evolution. And to your point, neither one of those were fully materialized. We continue to benefit from relatively stable macro and business trends from the quarter. In terms of our Cloud performance in Q4, paid seat expansion, cross- sell and migrations were the primary drivers of the better-than-expected results with in-line performance across the other Cloud drivers. Notably, the rate of paid seat expansion in SMB and enterprise continued to demonstrate quarter-to- quarter stability. In Data Center revenue was in line with our guidance with growth driven by pricing, and that was partially offset by strong migrations to cloud. Data Center growth also benefited from the renewal of several large customer agreements in the quarter. And then in marketplace, just to wrap it up. The outperformance was driven primarily by third-party app sales from Data Center. And then finally, I'd also highlight that we saw steady performance broadly across products, customer segments and regions. So all in all, Michael, really solid quarter overall, and we feel really good about how we ended the fiscal year.
3. TWC 升级后 seat 数不变，"多两位数"部分是机械对齐 → 冲击 E01/E04
CEO 在 Investor Day 说升级 TWC 后 Jira、Confluence、Loom 的 seat 数是一样的，多的是 AI credits。所谓 TWC 客户 seat 多两位数，CEO 自己说部分是打包要求几个产品 seat 数对齐。这是提价，不是更多人在用。
And when you look at the Teamwork Collection, which is our primary AI monetization vehicle right now, again, north of 1,000 customers, 1 million seats inside the first 6 months of upgrades of existing customers, some new in there, but mainly upgrades. choosing to upgrade Y, you get roughly 10x the AI or Rovo credits when you upgrade. We have a lot of other economic advantages. And I would argue, MAU advantages for the customer, you get the same number of Confluence, Jira and Loom seats, which increases your collaboration broadly, but again, far more AI credits. TWC customers use twice as many who use -- consume twice as many Rovo credits per paid user as non-Teamwork collection customers, and they have twice as many active agents. Now they have more credits. So like, yes, well, they've got more credits, they've opted in.
Ittai, if I can just chime in at sort of a nonfinancial and high level. Look, we're very clear that the system of work is about a continued growth of Atlassian into the knowledge worker population, right? When we talk about unleashing the potential of every team, our mission for over 2 decades, I would stress the every team part. We've got the software collection, which does very, very well. The DX and a lot of other things encompass and a bit like a pipeline, like there's a lot of great areas of that software collection. We've also got the Service Collection. So we're seeing great growth in Service Collection across HR teams, Finance teams, other areas outside of its traditional market in IT and Operations teams, but that's certainly an area of growth for us. So there's HR and Finance teams, we're seeing a lot in the Service Collection, for example, and you've seen us ship a lot of features. We also launched Customer Service this quarter as an application within that to go after a new set of teams we haven't been able to get to as well, you would say. In the Core Teamwork Collection, look, as we said in our letter, we were seeing double-digit seat expansion, more than double-digit seat expansion compared to people who buy the stand-alone applications. And that's baked into the Teamwork Collection packaging, but it's also because of the AI offering and how it works, right, in terms of getting -- you get -- you equate your Loom Confluence and Jira seats along with your additions, which often gains expansion. But given the nature of our applications and our continued growth in business teams across an Enterprise sales, Marketing, HR, Finance, et cetera, what that licensing structure allows those teams to do is to collaborate more. And collaboration is a very sticky and kind of viral activity. So that's where you're seeing that expansion coming through in our MAU and in our AI MAU and also ultimately ending up in our NRR and RPO numbers in terms of long-term commitments from customers that, that is where they're seeing expansion across business teams of all struts.
4. AI 工作量涨得比人快得多 → 冲击 E03
MCP 调用一个季度涨 400%，人用网页和手机的量几乎没变。这句可以理解为在说服务成本，但字面上说明 AI 带来的工作量不需要更多人登录。
Yes. Look, Karl, I can add on from the gross margin in the COGS side. I would say it's a classic Atlassian story of being thoughtful and long-term thinking -- long-term orientation. We've talked for 3 years now around our AI gateway and how we think about AI. We've been blending multiple models for a multiyear period, and we're getting really, really good at it. It's a nontrivial challenge. I'll give a huge shout out to all of our R&D teams who have been working on that cost to serve question for a long time in terms of getting the best quality at the best price and the best speed. People want [ strike ] fast answers from their AI across Teamwork Graph, across our enterprise search architecture, which is now the best in the world in terms of giving you answers across your knowledge estate, whether that's to a person or human or whether that's to an agent or something clearing in that manner from an AI perspective. Our ability to blend models, our ability to manage traffic costs over time is truly phenomenal. We get better at it every quarter. And you can see that we had a huge continued growth in AI usage. Again, Rovo assisted actions are up 50% quarter-on-quarter. And we've continued to do that and focus on the durable profitable growth that we've talked about for a while now. So it's a huge credit to everyone in the Atlassian team to be able to do that. It's a lot of hard work. It's a lot of long-term thinking about the architecture that we built underneath our AI to be able to give those answers in a -- in an incredibly competitive value way to our customers as well. So not only do we give them the best answers, we give them the best answers at an incredible price, which is part of the reason you're seeing the growth in the service question, the growth in the Teamwork Collection, again, as James mentioned service question, again accelerating AI being a huge driver of that service space. Again, hearing record numbers for customer conversions from legacy service providers, et cetera. So -- and you see that with our MCP calls as well as an example, that's also nontrivial to serve cheaply to manage that over time. MCP calls were up 400% in the quarter with barely a blip in people actually using the applications through the web or through a mobile client. So really happy where we sit from a technological perspective of being able to deliver this and more improvements to come.
5. 渠道商：agent 用一个 service account 就能操作 Jira → 冲击 E03
同一家渠道商一边说今年会卖更多 seat，一边说长期不需要每人一个 license，并预期 2-3 年后开发者大规模裁员；Service Collection 卖得好是因为 JSM 不再单卖，客户续约被强制转入。
SD: Yeah, and basically this is all because of AI and the more that everyone's using AI native coding tools or coding assistants or IDEs, there's really no need for everyone to have a license of Jira or everyone to have a license of any tool when you just plug the AI tool into like an MCP server that's talking to all these tools natively. It can all work as a single service account. There's really no need for multiple seats. So then, we need to monetize our products differently and it's like how much activity is Jira seeing and let's charge for that versus a per-user count and pricing it correctly so we don't lose revenue going forward when we switch from a seat-based to a consumption-based, unless there's some third pricing model that people can come up with that helps keep these companies afloat. I am concerned with all these companies, not necessarily just Atlassian, with, okay, we convert to this type of model to keep revenue flowing, but at some point AI is just going to take this functionality in-house and we won't need AI to manipulate Jira. We'll just do it internally inside the AI tool. And so at that point, why do we need these software tool vendors? I think it's going to be a struggle for them to convince us that they're still relevant. And if AI doesn't go that direction, if they just focus on continuing to improve, get rid of the hallucinations and wrong answers and get even... and we're seeing it go in that direction. It's getting better and better weekly. If they just focus on that and they leave the tool vendors alone, the tool vendors will survive. If we are in a world where the tool vendors are just partnering with AI and they're just getting better and better at input and output via the AI coding tools. Then they can survive if the Anthropics and the OpenAIs decide, yeah, we don't really need you guys. And there's no technical reason why something like a Jira has to exist once AI gets to that point. So they could go through the entire software development lifecycle set of tools, all the DevOps tools and just replace them one by one with internal features to the AI tool.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满
SD: Yeah, so the collections, the service collection is really the only one that is gaining traction, and the main reason it's gaining traction is it's a forced buy. So if you're renewing your JSM, you have to renew it into the collection because JSM is no longer available on its own, it is only available from the service collection. So I don't know if they're going to do that pattern for the other collections. We haven't seen that yet, but I wouldn't be surprised if they did, if you could only get Jira in the other collection now, which name escapes me, maybe the software collection.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满
6. Atlassian 自己：开发者效率提升约 25%，同时放缓招聘 → 冲击 E03
这正是市场担心客户会做的事，公司自己先做了。
Again, a lot of it is just thinking about how do we reinvent certain workflows to utilize and leverage AI. I think, like one of the interesting ones is we get asked a lot about our R&D team, right? Are you using some of these co-gen tools? I think what people miss is coding is only about 30%. We do surveys. You can look at third-party surveys. Coding's only 30% of the developer's time today, hands on keyboard development time coding time. So where we try to apply our AI capabilities is on that other 70% as well. Like how can you speed up some of the other 70% of what they do in collaborating, how do you figure out what we're trying to build? On the whole, we use cursor. We use cloud. We use a lot of tools, Rovo dev in the mix. Kind of the net-net of that is our developer is about 25% more efficient, which I think shows there's a lot of opportunity.
7. CEO 承认 JSM 同客户口径下有效率化 → 冲击 E01
被问到同店口径 seat 是否萎缩，CEO 承认部分客户在用 AI 提效，被 HR、财务等新场景抵消。
Sure. Thanks, Keith. Great question. Love questions about Service Collection doing fantastically. Look, we gave some stats in our shareholder letter, right? We passed 65,000 customers, which is a big milestone, 50% of the Fortune 500 as a business in and out itself and the enterprise side of that world growing over 60% year-on-year. So hopefully, from those sort of high-level statistics, Service collection is doing very, very well. It is definitely our fastest-growing product at significant scale. And that's a really important milestone for that business. As I said earlier, yes, that growth is happening on a like-for-like customer base like your sort of same- store sale analogy, I get what you're asking there. We're certainly seeing efficiencies coming in some of those customers. At the same time, as a challenger brand, we are seeing great growth in HR. We shipped to 12 months ago, a whole series of HR service management blueprints and other areas. We're growing really well in that sort of part of helping operate a business. Same in finance, same in other areas of operations, often like workplace management, these types of things. Service collection is doing very, very well in. And we feel very confident that we have a lot more growth to go get there. Secondly, on the asset management side, you've seen us take assets out of the service collection and put them into the core platform. That continues to be a big growth driver for us as we have a far more modern CMDB-like system as a graph compared to a lot of the legacy competitors. And as we connect the assets graph that you have of physical objects often to the Teamwork Graph, we're seeing our agents and our AI capabilities get significantly more powerful, and we're seeing great growth there. As such, in the last 6 months, more than 40% of the Agentic workflows that have been built are actually in Service Collection customers and Service Workflows. It's a very natural area to deploy agents -- AI agents into your service workflows to help improve the human agentic experience or the human agent experience of delivering value to the customer or the customer just getting the value directly themselves. So that is going very, very well. More than 2/3 of our service collection customers are using it for non-IT use cases at the moment, which is a great sign that, that is happening. Two other things maybe, one, you see we're a leader in our enterprise service management wave. The analyst community continues to recognize us as a leader and a visionary, but also a significant challenger and growth brand. And we continue to see a lot of migration from legacy service management platforms on to service collection for much higher ROI, much better cost equation with a much more modern stack and user experience, and that's really great for us. Lastly, and it shouldn't be last, it's definitely not the most minor. We only GA our customer service management app inside the last quarter. So that's delivering great efficiency results for us in running parts of our customer service, as we've said, and very early in that journey, but really excited about how that can continue to grow the Service Collection as it continues to power a large part of the license growth.
8. 2025 年 10 月：云指引上调只反映迁移，有机增长指引反而下调 → 冲击 E05
Yes. Great question. Thanks. And I'll try and clarify. So we are raising our cloud revenue outlook by 1.5 points to 22.5% year-over-year. That is only to reflect the stronger migrations performance and the outperformance in Q1. So we now expect migrations to make a mid- to high single-digit contribution to cloud revenue growth in FY '26. And for that migration upside to land in the back half of the year, just given the data center expiration base. And to your question directly, it's important to note we haven't made any changes to our other organic drivers of cloud revenue growth in our guidance. So we continue to maintain a conservative and risk-adjusted approach on all those other variables in the cloud and from a cloud revenue growth driver perspective for the rest of the year.
Keith, this is Joe. I'll go first, and Mike will follow on with additional context. I'd say the key development that we saw in Q1 was that we had significantly stronger-than-expected cloud migrations from data center, which is a great thing for our business. As you know, this has been a big area of investment for us. It's a key strategic priority. It allows us the opportunity to provide more value to customers in ways that we simply can't on data center, and we can offer capabilities such as automation, analytics and AI, and so basically, it's definitely the most valuable and secure experience we can offer to our customers. So it's a great thing for the business. Cloud migrations also, however, have an impact on the timing of revenue recognition because cloud revenue is recognized ratably and data center has a combination of upfront recognition and some ratable. And then lastly, the move to cloud also impacts marketplace revenue because we have a lower take rate on cloud app sales than we do on data center apps. All other organic growth drivers in our business in Q1 were either slightly better or in line with our expectations. And we've maintained that same guidance approach on those factors that we held 3 months ago. So we continue to hold a very conservative and risk-adjusted outlook for all the other variables in the growth equation outside of migrations. So with that Q1 performance and momentum, we've adjusted our full year outlook for a greater cloud migration forecast. And we've left all other organic driver assumptions in place. And because of those revenue recognition timing differences between cloud and data center and the impact to Marketplace, this drives the 0.5 point decline in our organic revenue growth outlook for the rest of the year. And so that's the math underneath the guidance that we're giving for the full year on the organic part of the business. Mike?
9. 卖方和渠道：预算降优先级、Linear 抢份额、SMB 裁员 seat 掉 20%
这三条都是二手转述或个例，量级不能外推，但方向一致：小客户和开发者重度场景是 seat 最脆弱的地方。
UBS says recent enterprise software checks point to worsening IT budget conditions for SaaS vendors, with customers increasingly prioritizing AI, cloud infrastructure, cybersecurity, and data spending over traditional seat-based application software. The firm highlights Workday, Salesforce, Atlassian, and HubSpot as among the most frequently deprioritized categories, while shorter contract durations, pricing pushback, and seat optimization are emerging as new headwinds to cRPO/bookings growth. UBS continues to favor AI/data infrastructure beneficiaries like PLTR and SNOW, arguing “build vs buy” dynamics and internal AI application development are creating additional structural pressure on broader SaaS spending.
来源背景：UBS 卖方（TMTB 转述）
Citizens maintained a Market Perform rating after channel checks, with 11 of 18 key data points coming in negative, led by evidence that Jira is losing share to Linear while the wind-down of Atlassian’s Data Center business remains the largest structural headwind. The firm highlighted encouraging momentum in Rovo, where AI usage is growing 20%+ m/m, and believes Collections remains an attractive monetization opportunity, but FY27/FY28 EPS estimates remain below Street. Citizens argues the stock’s 2.6x CY27 EV/revenue valuation appropriately reflects slower growth and competitive pressures.
来源背景：Citizens 卖方（TMTB 转述）
Now they've gone from 50 to 40 people because they laid off a bunch of folks. All of a sudden, there's a decline of roughly 20%.
来源背景：前 Atlassian Global Head of Enterprise Sales（In Practise 推文引述）
Thesis 证据分析
按生意上的问题组织。每节先给判断，再列关键原文，最后可以展开该节全部原文。绿色是支持 thesis 的命题，红色是反对，灰色是口径不齐或中性。
Seats 还在涨吗
判断：在涨，没有萎缩；但没有一个数字，速度是"稳定"而不是"加速"。
2025 年 5 月到 2026 年 9 月，管理层在每一场业绩会和卖方会议上都说 seat 在扩张，而且越说越强：从 2025 年的"very consistent""absolute seat growth"，到 2026 年 Q3、Q4 两次把云收入超预期直接归因于"cross-sell 和 seat expansion"。CRO 说续约时没有看到 seat 减少。Investor Day 上 CEO 放了一张"剔除迁移的云 seats"曲线，让大家"拿尺子量"，但没给任何刻度。
能说明速度的只有 CFO 的"稳定 4-6 个季度"。外部证据也是如此：渠道商 Q1 看到 seat 大涨，但几乎全来自一个国防客户 (300 → 5,000，专家自己怀疑对方配不齐人)，非国防业务"平淡、符合预期"。
Customers aren't just buying in, they keep expanding. So let me give you an interesting chart. I think it will be interesting for you. Let's look at total cloud seats, excluding all migrations. No migrations involved. And despite the, oh, seat-based, it's going to disappear. We can get out your rulers if you want, I know you do. You can see visually, it's not slowing down. It's a pretty amazing curve. Again, this is excluding all migrations, cloud seats. They continue to grow every quarter, every year, durable growth as we talk about.
Thanks, Gregg. I can start with that and maybe throw it over to Mike. To Ryan's question a little bit earlier about some of the cloud outperformance drivers that we saw. I talked about the cross-sell side of it, and we are seeing customers continue to go wall to wall with Atlassian here and continue to adopt higher additions and move towards Teamwork Collection, Service Collection. But I think importantly, we're seeing that growth in Teamwork Collection while also seeing continued strong seat expansion in our core Jira and Confluence offerings here. And this expansion is broad-based. Recall that roughly 2/3 of users across Jira and Confluence are knowledge workers, right? So these are non-software development teams, really sitting in HR, legal, finance and really speaks to that diversified user base that we're seeing here. And I think this growth does demonstrate that in AI-driven world, the need for collaboration at the enterprise scale, tracking, planning, managing work, it's actually increasing. And Atlassian is a mission-critical platform here, helping our customers orchestrate their teams, their agents, their workflows to really unlock that value in AI and driving real ROI and real outcomes here. And that really was the other primary driver of the outperformance that we saw in cloud, including the cross-sell that I talked about a little bit earlier.
Sure. So I would say, firstly, we're not seeing a reduction in the number of seats, right? The conversations that we're having with customers are not around that. We're not seeing that in the conversation. We're not seeing that at the time of renewal with our customers. Instead, the conversations that we're having with our customers are very different. I'll give you an example. Just earlier today, I spoke with a large U.S.-based consumer tech company that was talking to us about a long 7-year term agreement that they wanted to enter into us, and this is a company that has over 100,000 seats with us both in the developer and the IT space. We firmly believe that having the ability for developers and nondevelopers to create software leads to more software. And same as what Aaron said, this means that there will be, one, the hiring of more developers and the creation of more software. And more software leads to more projects. What we have found, in fact, is that we see -- and for companies that are utilizing these cogeneration tools, there actually is a 5% increase for us in seats. And a 5% increase in fees obviously leads to expansion with our customers. And I know that's contrary to the headlines that you just gave there, but this is having a tailwind for us rather than a headwind. And that is what is leading to the headlines that we made around $1 billion -- our first $1 billion quarter in Q2, the 600 customers that are now paying us over $1 million and an RPO growth that is north of 40%. If we were facing all of these headwinds, we wouldn't be able to deliver numbers consistently like this quarter-over-quarter. So if anything, like I said at the beginning, AI is actually a tailwind rather than a headwind for us.
Yes. Thanks for the question, Ittai. In terms of where we see the expansion, I'd say it's broad-based. It's across both tech and non-tech users. We are making a lot of progress on what we call non-tech or business users, particularly with the TWC product. Those seat expansion rates, both across the enterprise and SMB remain stable. That's been the case for 4 to 6 quarters now. So we feel really good about the continued progress on that. And that's the way I would describe from a paid seat expansion perspective, sort of the color in terms of what we're seeing on that front.
SD: Yeah, so the first quarter was going smoothly until mid-March. And then mid-March, it exploded for us, and we saw a huge amount of seat growth, and primarily it was because of our main vertical. So our main vertical is PubSec and defense. And so towards the middle to end of March, the activities in the Middle East have spawned a lot of projects to accelerate. And the main way that they're accelerating is increasing developer headcount. We have one particular client that I can't mention went from a 300-seat count to a 5,000-seat count pretty much overnight.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满
SD: Now, I have a hard time believing they'll actually staff that. But they requested that many licenses.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满
Yes, Keith, thanks for the question. So on the data center side, the Q3 revenue beat, as we mentioned, was primarily driven by recognizing greater-than-expected upfront term license revenue within that quarter. Since our announcement of the data center end of life back in September, we've had a couple of quarters now to really better understand some of the signals that we're seeing from our customers in terms of their buying behaviors, especially Q3 being the largest expiry base for us. So let me unpack that a little bit more for us here. So first, I would say that the migration to the cloud is on track and continues as expected. We're pleased with what we're seeing there. We still expect that to contribute mid- to high single digits on cloud growth. And second is that the retention rates on our data center business remain incredibly robust, in fact, actually outperformed expectations in the quarter. And third, for some of our largest customers with more complex migration, they remain committed to transitioning to the cloud. But it's going to be a multiyear journey for them, right? They've got a lot of deep customizations, change management. It's going to take these customers time, right? Often many of these have tens of thousands of users, some with over 100,000 users. So this category of customers, we saw a pull forward of purchasing and expansion activity into Q3 from future periods. And we also had a pricing change in March. That further catalyzed this dynamic. So as a result, that drove greater-than-expected upfront term license revenue recognized in the quarter. In fact, relative to our expectations for Q3, we recognized approximately $50 million more in upfront term license revenue. And that's some of the trends that we've been seeing since our announcement of end of life back in September, but more pronounced in Q3 given the size of the expiry base and the pricing catalysts that I mentioned. And maybe lastly, the cohort of DC customers that are actively planning and transitioning to the cloud, we're seeing these customers moderate their seat expansion. versus historical trends we've seen. Again, I'll mention that retention rate remains incredibly high, but now expecting a more muted level of data center expansion from these customers going forward as they try to move to the cloud. We're still seeing really nice uplift when customers move from DC to cloud. So net-net, what we're seeing is that our largest strategic customers continue to deepen their commitment at lasting, whether that's on DC or cloud, and we're working hard to meet them where they are and help them accelerate that transition so they can unlock all the AI and agent capabilities in the cloud. So hopefully that gives you a little bit of color there, Keith. And maybe I'll just share that with these dynamics sort of playing out across the year on data center, with Q4 yet to play out where revenue rec is being pulled into FY '26 from FY '27, we recognize that there's lumpiness in that pull-forward effect in data center and having the timing of -- and that does impact the timing of reported revenue RPO and CRPO. So internally, of course, we look at a variety of metrics to performance manage our business, including a really healthy ARR. So next week, we're going to be holding that investor forum at our Team '26 Conference and to help guide investors through the revenue recognition timing dynamics on the data center side. We'll look to enhance our disclosures and share historical subscription ARR, which will help normalize some of those timing effects and help everyone better understand the underlying strength of the overall business. So all up, we feel really good about our execution, the runway that we still have ahead of us. So more to share next week at the TEAM event and the investor forum.
展开全部 46 条原文 (按时间)
- 支持 FQ3 云收入超预期的两大来源之一是好于预期的 paid seat expansion（未量化）
Yes. Thanks for the question. From a Cloud perspective, we did deliver 25% year-over-year revenue growth. That was better than we expected coming into the quarter. The core underlying performance in our Cloud business was very similar to Q2 with trends from Q2 continuing into Q3. The variance to our expectations was driven by two factors. It was better-than-expected paid seat expansion and Data Center migrations. All the other drivers in this part of our business, whether it's cross-sell of additional products or adoption of higher-value additions or top of funnel performance or customer retention, those were all in line with our expectations going into the quarter. In terms of looking at the level of beat relative to the guidance in Q3 versus first half, I would just start by saying that in both Q1 and Q2, those were exceptionally strong quarters where we executed really well against a very robust set of opportunities in pipeline. Then I think fundamentally, Q3 is simply a seasonally slower quarter for us. So there was less opportunity for the level of outperformance you saw in the first half of the year. And then I know not specific to Cloud but more generally, there were some mechanical dynamics, particularly in Data Center and Marketplace that negatively impacted revenue growth on the margin. And I think taking all of this into account helps explain most of the difference between our H1 performance to guidance in Q3.
- 支持 管理层认为有巨大的用户扩张空间（版本 + 更多知识工作者 wall-to-wall），属愿景未量化
Sales of premium enterprise editions are up over 40% year-on-year. And the cloud customers who have deployed Rovo agents to now tens of thousands of unique workflows connected to automation are seeing great value. So I would say that from that position of strength, we feel very good about the capabilities that we've built and we want to get them into as many hands as possible and as fast as we can amongst our user base, right? We believe that by maximizing usage and adoption, we -- yes, we've always been patient with monetization, but we think we have a massive user expansion opportunity, both in terms of additions and also growing wall-to-wall with more knowledge workers in those organizations. So that fits really well with the way that we think about building software in general. Again, we're always very thoughtful about balancing growth and capital usage. As we mentioned at the Investor Day, we've done some amazing work in R&D to make those capabilities able to be deployed at a reasonable cost, which is obviously a gateway in order to be able to make this sort of a move.
- 支持 上云客户的扩张（加 seats）和交叉销售都比本地部署更强
Sure. So Atlassian historically had a server business. Think of this as a perpetual license model. So we would basically have customers buy the server software and really only pay us a maintenance fee there on. So several years ago, we decided to switch to a cloud model for two reasons. One, technologically cloud would allow us to offer innovation that server technology would not; and second, we wanted to shift to a subscription model from the perpetual license model. And we began this journey in 2020 and we are in the -- I would describe it as the final leg of that journey. And we've been phenomenally successful with that transition. So we've done better than what we had internally modeled in terms of customer churn. We've done better in terms of what we expected expansion to look like. So the fundamental thing to understand about the business is that the shift from perpetual to subscription business is complete for us in terms of we have -- we no longer have a perpetual license model, so the end of life the server product and data center and cloud both have a subscription model. And customers who adopt our cloud products, like I said before, show higher degrees of expansion that is adding more seats as well as higher degrees of cross-sell, adding more products. And we have a lot more cloud products than we do server products naturally because server was the legacy business. And over the years, we've managed to successfully migrate a lot of our server customers. So we went from the business being entirely server in DC, well over 90% of our customers now around cloud and 90% of our customers now choose cloud. And the set of customers, I would describe our journey as the leg that is about to come is that we have a set of enterprise customers that are some of our largest and most complex customers that have regulatory rules, that have constraints that need them to run hybrid environments, those are the customers that we're focused on over the next couple of years to bring them over to the cloud platform. I would describe the migration overall as having been extremely successful and better than what we expected.
- 支持 过去 4 个季度 seat 增长很稳定、很健康（没给数字）
I think what's also interesting, when we look at our seat growth, our seat growth has been very consistent over the past 4 quarters. So we continue to see really healthy seat growth. What's been also really interesting is when you cut it by tech startups, right? You would imagine that tech start-ups tend to be the most tech-forward companies. Their seat expansion patterns have been consistent with the seat expansion patterns that you see with SMBs across other industries. So that's a really encouraging sign for me, which again, I think, shows you our ability to serve not just probably developers, but also other types of users. And then earlier, you mentioned our focus on enterprise, right? And so while we certainly drive a lot of our growth via seat expansion, we also have developed different growth vectors in terms of upgrading people to premium and enterprise additions upselling and focusing on these enterprises, cross-selling across the broader portfolio of products and then pricing, right, capturing it via consumption pricing or perhaps per seat pricing over time. So I think we feel really good about kind of the culmination of these different vectors and our ability to drive durable growth no matter how the world continues to evolve with Gen AI.
- 支持 seat 扩张模式没有偏离，继续看到健康的、绝对值的 seat 增长
Yes. And again, I think it's important to remember that fundamentally Atlassian is a collaboration company, right? Like all of our products facilitate collaboration. And so as the output increases of software because no doubt it will, there become all these new problems that humans have to collaborate together on and solve and Brian echoed all of that, right? Like how do you manage all this new software? How do you sell it? How do you get feedback from what customers want so that you can iterate on that software? Again, there's an abundance of new problems that get created. I think it's also important to remember that already today, 50% of our seats are technical in nature, so that's developers, engineering, IT. And already, we serve 50% that are nontechnical. So those business users, and we'll continue to push more in that direction. When you actually look at some of these SMB tech start-ups that you would expect to be much more kind of embracing or forward tech embracing, when you look at their patterns of seat expansion, it's been consistent for the past year. So that's good. And then also when you contrast that against SMBs of similar size in different industries, it's very consistent. So you don't see any deviation in kind of seat expansion patterns. Continue to see very healthy seat growth, continue to see absolute seat growth. I think the other interesting anecdote is when we look at some of these CEOs or people that make kind of bombastic proclamations, when you look at the underlying customer data, we actually see seat expansion within a lot of these companies. And so it's -- I think we feel really good as we continue to see seat growth because of all these initiatives and projects that continue to develop, as Brian said, but also as we kind of continue to push more to serving these nontechnical users.
- 支持 Q4 云超预期的主要来源是 paid seat expansion、cross-sell 和迁移
On Cloud revenue guidance coming into the quarter, we'd assume some negative impact on paid seat expansion rates and cloud migrations from macro uncertainty and our Enterprise go-to-market evolution. And to your point, neither one of those were fully materialized. We continue to benefit from relatively stable macro and business trends from the quarter. In terms of our Cloud performance in Q4, paid seat expansion, cross- sell and migrations were the primary drivers of the better-than-expected results with in-line performance across the other Cloud drivers. Notably, the rate of paid seat expansion in SMB and enterprise continued to demonstrate quarter-to- quarter stability. In Data Center revenue was in line with our guidance with growth driven by pricing, and that was partially offset by strong migrations to cloud. Data Center growth also benefited from the renewal of several large customer agreements in the quarter.
- 口径不齐 SMB 和企业的 paid seat expansion 速率环比稳定：说明没恶化，但也不是加速；未给数字
And to your point, neither one of those were fully materialized. We continue to benefit from relatively stable macro and business trends from the quarter. In terms of our Cloud performance in Q4, paid seat expansion, cross- sell and migrations were the primary drivers of the better-than-expected results with in-line performance across the other Cloud drivers. Notably, the rate of paid seat expansion in SMB and enterprise continued to demonstrate quarter-to- quarter stability. In Data Center revenue was in line with our guidance with growth driven by pricing, and that was partially offset by strong migrations to cloud. Data Center growth also benefited from the renewal of several large customer agreements in the quarter. And then in marketplace, just to wrap it up.
- 支持 上季度迁移升级的 seats 数同比翻倍（迁移口径，不是新增 seat）
Thanks, Kash. Look, I always appreciate the applause, I guess. Let me say a few things about Ascend and cloud migrations. Firstly, the partner and customer reaction has been fantastic. I think that's because we have well telegraphed to our customers, our partners, the entire community, that cloud is the future. It's the best experience for our customers. So we had very little surprise. We're very thoughtful and long term about how we've managed the transition, I think, and the reaction has been very positive as a result. I think the removal of technical barriers and the delivery of innovation in the cloud has been a combination of effects that's really bringing that through. You see that in the number of customers that mentioned to me, for example, AI is one of the big reasons that they are moving to the cloud. We've learned an awful lot about how to help those customers manage that upgrade through the fast shift program, through our amazing partner network, through a lot of different things over the last 5-plus years. And you see that in doubling the number of migrations -- seats that were upgraded in the last quarter that doubled year- on-year. That's a huge achievement for us. And it is thanks to those partners that you talked about and everybody else, and you see it in us raising the cloud revenue outlook. So I think cloud is ready for our customers. You see that in FedRAMP Moderate in government cloud, isolated cloud, multi-cloud strategy. So we feel incredibly bullish that we have the experience to do this. We feel that it's the right experience for our customers, and we are thoughtfully managing that migration as you pointed out. So I don't think the playbook has necessarily changed as much as every year and every quarter that goes by, we get better at it. We did lots and lots of large migrations last quarter across different geographies, different industries, pretty much any country that I go to, any city, I can point to a large customer in that geography in their industry that has already moved. So that's what gives us a lot of that confidence going forward, and allows us to reiterate those long-term targets that we laid out.
- 支持 CEO 称 120% NER 等扩张率没有变化，paid seat expansion rates 体现云业务强劲（未量化）
I can probably jump on DJ, just to say 1 or 2 things, if I might. Firstly, it's worth reiterating that our expansion rates 120% NER, et cetera, aren't changing. So when Joe says, we are continuing with our cloud, the guidance in other areas I think those are really strong numbers, and we should reiterate that. We feel great strength in the cloud, right? Teamwork collection going very, very well, only 2 quarters in. Our AI delivering our enterprise platform. All of these things lead to a very strong cloud business in and of itself that continues to grow. And the Ascend program and migrations are additive to that, which is really great. When I talk to our customers that are already in the cloud at scale, they are bullish about their continued adoption of more apps and collections of more areas that they will move to Atlassian. And you can see that showing up in both our paid seat expansion rates, our cloud growth rate in and of itself, our RPO growth rate and our recommitment to our 3-year 20% CAGR that we gave out. So incredibly bullish about the cloud business as a whole as a result of AI enterprise and the system of work. All the things that we've been saying for a while now continue to come due with our customers. And I'll tell you, having spent a lot of time with them. they're all incredibly excited about what we are delivering to them every single day, and it's a credit to the entire Atlassian team.
- 支持 云上大客户在采用更多产品，体现在 paid seat expansion rates 中（未量化）
I can probably jump on DJ, just to say 1 or 2 things, if I might. Firstly, it's worth reiterating that our expansion rates 120% NER, et cetera, aren't changing. So when Joe says, we are continuing with our cloud, the guidance in other areas I think those are really strong numbers, and we should reiterate that. We feel great strength in the cloud, right? Teamwork collection going very, very well, only 2 quarters in. Our AI delivering our enterprise platform. All of these things lead to a very strong cloud business in and of itself that continues to grow. And the Ascend program and migrations are additive to that, which is really great. When I talk to our customers that are already in the cloud at scale, they are bullish about their continued adoption of more apps and collections of more areas that they will move to Atlassian. And you can see that showing up in both our paid seat expansion rates, our cloud growth rate in and of itself, our RPO growth rate and our recommitment to our 3-year 20% CAGR that we gave out. So incredibly bullish about the cloud business as a whole as a result of AI enterprise and the system of work. All the things that we've been saying for a while now continue to come due with our customers. And I'll tell you, having spent a lot of time with them. they're all incredibly excited about what we are delivering to them every single day, and it's a credit to the entire Atlassian team.
- 支持 Seat 扩张跨技术和非技术用户；企业和 SMB 的扩张率稳定了 4-6 个季度（稳定，非加速；未拆分新客户/存量）
Yes. Thanks for the question, Ittai. In terms of where we see the expansion, I'd say it's broad-based. It's across both tech and non-tech users. We are making a lot of progress on what we call non-tech or business users, particularly with the TWC product. Those seat expansion rates, both across the enterprise and SMB remain stable. That's been the case for 4 to 6 quarters now. So we feel really good about the continued progress on that. And that's the way I would describe from a paid seat expansion perspective, sort of the color in terms of what we're seeing on that front.
- 反对 JSM 同客户可比口径下：部分客户出现效率化（agent 替代），被 HR、财务等新场景的增长抵消
So hopefully, from those sort of high-level statistics, Service collection is doing very, very well. It is definitely our fastest-growing product at significant scale. And that's a really important milestone for that business. As I said earlier, yes, that growth is happening on a like-for-like customer base like your sort of same- store sale analogy, I get what you're asking there. We're certainly seeing efficiencies coming in some of those customers. At the same time, as a challenger brand, we are seeing great growth in HR. We shipped to 12 months ago, a whole series of HR service management blueprints and other areas. We're growing really well in that sort of part of helping operate a business.
- 支持 多年期大单涉及数万个 seats
That, in return, makes those customer conversations about longer-term commitments, right? As you pointed out, you see that in our numbers, right? Our RPO at 44%, growing -- accelerating for the third quarter in a row is a really fantastic vote of confidence, I believe, from those customers, right? Those are tens of thousands of seats signing multiyear deals, that are voting on not the platform for 2026 for them in AI, but the platform in '27, '28 and '29. And those customers are seeing what we're doing, seeing our progress and voting with their feet. There's a lot, obviously, of noise out there in the market, right? As I said in my shareholder letter, there's no doubt about that. But when I talk to customers, they believe that we're helping them through a lot of that noise. We're delivering for their software teams, their business teams and their business process, and they're able to get efficiencies and improvements today. And I would say they want all the same things they've wanted in the past. Sometimes in these type of times when there's a lot of noise, right, we can forget the fundamentals, right? Enterprise customers want a platform they can trust. They need it to be compliant and secure and all the things they've always needed. They want great ROI. They want efficiency in their business.
- 支持 客户对话和续约时都没看到 seat 减少
Sure. So I would say, firstly, we're not seeing a reduction in the number of seats, right? The conversations that we're having with customers are not around that. We're not seeing that in the conversation. We're not seeing that at the time of renewal with our customers. Instead, the conversations that we're having with our customers are very different. I'll give you an example. Just earlier today, I spoke with a large U.S.-based consumer tech company that was talking to us about a long 7-year term agreement that they wanted to enter into us, and this is a company that has over 100,000 seats with us both in the developer and the IT space. We firmly believe that having the ability for developers and nondevelopers to create software leads to more software. And same as what Aaron said, this means that there will be, one, the hiring of more developers and the creation of more software. And more software leads to more projects. What we have found, in fact, is that we see -- and for companies that are utilizing these cogeneration tools, there actually is a 5% increase for us in seats. And a 5% increase in fees obviously leads to expansion with our customers. And I know that's contrary to the headlines that you just gave there, but this is having a tailwind for us rather than a headwind. And that is what is leading to the headlines that we made around $1 billion -- our first $1 billion quarter in Q2, the 600 customers that are now paying us over $1 million and an RPO growth that is north of 40%. If we were facing all of these headwinds, we wouldn't be able to deliver numbers consistently like this quarter-over-quarter. So if anything, like I said at the beginning, AI is actually a tailwind rather than a headwind for us.
- 支持 半导体客户 4 万 seats 上线后 10 天再扩 8,000（20%）
Yes. And so as you said, we have 5 million monthly active users at the end of Q2. We took a very different approach than other companies. We chose not to sell Rovo. Instead, we said that any customer who is moving to the cloud with us will get Rovo, specifically any customer that moves to Teamwork Collection, which is what I call our hero motion. Teamwork Collection includes Jira Confluence, Rovo and Loom. They get 10x the Rovo credit. We have been very successful with Teamwork Collection. It is really propelling our -- overall, our Rovo revenue stream, let's say. We have over 1 million seats in the cloud, and we have 1,000 customers that have moved to Teamwork Collection since April of last year. That is beyond what we expected. And we are very happy with the traction that we have received there. Specifically, what we have experienced is customers who have moved to Teamwork Collection who have embraced Rovo, we have seen a 20% increase in addition -- expansion as they have embraced Teamwork Collection, and we have seen an uptick in the Rovo usage as well. Just in the month of February, we had a large semiconductor company here on the West Coast go live with over 40,000 seats. Following their go-live 10 days later, there was an expansion to the tune of 8,000, the 20%. And that trend, we continue to see. We are consistently seeing more and more use cases emerging from our customers around the world. One thing that we are doing to drive more usage and more monetization is every one of our Teamwork Collection transactions that we were doing has a use case for Rovo embedded into it from our sales team, which means that our customer success team at the time of closure of that transaction will then pick up the use case and work with the customer and our forward deployed engineers that are embedded with the customer to help them activate the use case so that they can then get the journey started. And usually, what we see is that this leads to the creation of 3-plus agents at each of our customers.
- 支持 一家美国消费科技公司（10 万+ seats）在谈 7 年期合约
Sure. So I would say, firstly, we're not seeing a reduction in the number of seats, right? The conversations that we're having with customers are not around that. We're not seeing that in the conversation. We're not seeing that at the time of renewal with our customers. Instead, the conversations that we're having with our customers are very different. I'll give you an example. Just earlier today, I spoke with a large U.S.-based consumer tech company that was talking to us about a long 7-year term agreement that they wanted to enter into us, and this is a company that has over 100,000 seats with us both in the developer and the IT space. We firmly believe that having the ability for developers and nondevelopers to create software leads to more software. And same as what Aaron said, this means that there will be, one, the hiring of more developers and the creation of more software. And more software leads to more projects. What we have found, in fact, is that we see -- and for companies that are utilizing these cogeneration tools, there actually is a 5% increase for us in seats. And a 5% increase in fees obviously leads to expansion with our customers. And I know that's contrary to the headlines that you just gave there, but this is having a tailwind for us rather than a headwind. And that is what is leading to the headlines that we made around $1 billion -- our first $1 billion quarter in Q2, the 600 customers that are now paying us over $1 million and an RPO growth that is north of 40%. If we were facing all of these headwinds, we wouldn't be able to deliver numbers consistently like this quarter-over-quarter. So if anything, like I said at the beginning, AI is actually a tailwind rather than a headwind for us.
- 支持 技术用户和非技术用户两边都在增长
Yes. Look, I think we continue to grow with those technical users and nontechnical users. So across both vectors, which is great to see. I think -- two things I would first share. It's like one, we look at those customers that are using these co-gen tools. So again, trying to address some of the fears out there. When we look at those customers that use a cursor that use a cloud code via MCP across thousands of customers, so not an insignificant number. Those customers are driving 5% more tasks through Jira. It's leading to actually 5% greater monthly active usage because to your point, more people have to collaborate to manage this increased complexity or throughput of work. And then ultimately, it's driving those customers to expand their seats 5% faster than those customers that don't use these co-gen tools. So it's really interesting. And again, it kind of backs up what we've always believed in, but we thought it was important to share with the market. And then number two, I think we get questions about, okay, what about these hot new AI start-ups. So we picked the nonsubjective list. The Ford's AI 50. I didn't want to cherry pick some stats. I just drew that over to my data science team and said how many of these -- the hottest startup -- AI start-ups in the world are our customers today? 60% are already our customers today. And they're all growing their seat counts much faster than the overall corporate rate, which is probably not too surprising, but again, it's like -- so we feel good about our kind of position in an AI world, especially even with these hottest companies that investors may have questions about. Hey, do you know those companies work in a different way? It seems like we're well positioned even with these companies.
- 支持 企业客户在扩大 seat 数、签更大金额和更长期限的合约
In this moment In time, I think some people would say, okay, our enterprises, maybe taking a beat to evaluate their investment options or evaluate, kind of, the vendors. If anything, we're seeing a lot of these enterprises vote with their wallets in this moment in time. So we saw enterprises grow their seat counts. They sign for larger dollar amounts, and they're signing for longer contract duration which showing the commitment to the Atlassian platform, showing that they want a trusted vendor to run their Agentic workflows on and a reliable platform that has context behind the type of work that they're trying to do. So the enterprise efforts are going great. The actuary report card that you referenced is the result of three things, I think. We've talked to in more recent times about three company priorities. And so it's -- and serving enterprises delivering AI and what we call the Atlassian system of work, which is moving more towards this collection strategy or holistic solutions as opposed to individual point products. So those three strategic initiatives are coming together, and that's kind of the combination culminating in the report card that you just referenced and that's driving great results right now. We want to continue to push ahead on those three strategic priorities. And I think those are going to be important ones as we drive more durable growth into the future.
- 支持 升级 TWC 的客户 seat 扩张比单买 Jira 时多 10 个百分点以上
Yes. So the teamwork collection, as you talked about, has only been in the market for about 3 quarters now. As we rolled out Rovo, our AI capabilities across our platform, you get a certain number of AI credits with your standard subscription. With teamwork collection, you get 10x the amount of AI credits relative to U.S. stand-alone Jira stand-alone confluence subscription. So it's very appealing for customers as they think about their kind of AI journey because when you talk to our customers, they don't want consumption base pricing, at least today for our products. I think it makes sense for certain other models. But for our products, they don't really want to pay on a consumption basis. So today, if you use AI credits above and beyond your allotment on your standard Jira subscription or Confluent subscription, then you start to pay us on a consumption basis. But again, if you upgrade the Teamwork collection, you get 10x the amount of A credit. So it's a way to kind of forgo that worry or monitoring of AI credits today. That's been the #1 driver of why customers want to upgrade or choose to upgrade the Teamwork Collection. A lot of success early on, it's our expectations. 1,000 customers have upgraded to Teamwork collections. It equates to 1 million seats sold. Those customers that have upgraded Teamwork collection have expanded their seats versus their stand- alone footprint on Jira by double or by more than 10 percentage points of seats expansion, which is great to see that drives additional kind of attach of Confluence in Loom and then ultimately, the 10x amount of AI credits. So there's a lot of compelling kind of package within Teamwork Collection.
- 支持 Q1 末出现大量 seat 增长，主要来自 PubSec/国防，单一客户 300→5,000 seat
SD: Yeah, so the first quarter was going smoothly until mid-March. And then mid-March, it exploded for us, and we saw a huge amount of seat growth, and primarily it was because of our main vertical. So our main vertical is PubSec and defense. And so towards the middle to end of March, the activities in the Middle East have spawned a lot of projects to accelerate. And the main way that they're accelerating is increasing developer headcount. We have one particular client that I can't mention went from a 300-seat count to a 5,000-seat count pretty much overnight.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 口径不齐 5,000 seat 是申请量，专家怀疑客户能否真正配齐人员；且可持续性存疑（地缘驱动）
SD: Now, I have a hard time believing they'll actually staff that. But they requested that many licenses.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 口径不齐 PubSec 增长受地缘事件驱动，专家不确定可持续
SD: I don't know how sustainable this is going to be. Are we going to hear something from the administration that makes these projects go back to their scheduled programming? Or are we going to see this just continue to increase to other PubSec projects, you know, outside of the particular area of defense that just exploded on us? Exploding is a weird term to say for that particular... hopefully...
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 中性 非 PubSec 业务 Q1 平淡、符合预期（Q1 续约少）
SD: It was, you know, uneventful, as expected.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 支持 专家预期 Atlassian 今年会卖更多 seat；全年增长预期 30-35%（此前 20-25%）
SD: So bringing it back to what do I think for Atlassian sales, I think that we're going to be selling more seats. And, you know, depending on when that conversion is from seats to consumption. How does Atlassian take care of the partners when we do that? Because are we selling consumption packages or is Atlassian just have the gas meter on the different products and they're just raking it in without the partners? So that's...
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 支持 云超预期的两大驱动是 cross-sell 和 seat expansion（迁移符合预期，不是超预期来源）
Gregg, thanks for the question on that. Yes, on the cloud side, again, we saw strong performance reaccelerating to 29% year-on-year there. And maybe I'll start with the fact that DC migrations again to the cloud were in line with our expectations and not the realistic contributor to that $50 million beat on the print. It's progressing well, and we still expect that to contribute that mid- to single high-digit growth to cloud, like we mentioned. So the real 2 primary drivers that we talked about on the outperformance is really that cross-sell and seat expansion. . So on the cross-sell side, we saw outperformance in our collections business across Service Collection and in particular, Teamwork Collection. We have got Jira and Confluence, Loom and Rovo. And just keep in mind that right now, Teamwork Collection is really the best vehicle for our customers to buy and unlock AI and all the agent capabilities across the Atlassian platform, right? Customers are upgrading to TWC because of the increased AI credits. We're giving 10x more credits on Rovo versus the stand-alone subscription. And I'm sure Mike can touch a little bit more on some of the progress that we're seeing there. But importantly, we're also -- the other driver here is that we're seeing that growth in TWC while also seeing continued seat expansion in our core Jira stand-alone offering, right? And I think that really speaks to how an AI-driven agentic world, Jira and the Atlassian platform really remain core to enabling customers to manage their workflows and collaboration to fully unlock that value of AI. So whether it's TWC or standalone Jira offerings, we're launching a ton of new value that's really enabling our customers to deploy agents to do the work, to capture that agent activity alongside work history, full permissions and audit trails and admin governance. So a lot of great traction here, as you can see in our Q3 print.
- 支持 TWC 增长的同时，Jira 单品也在持续 seat 扩张（未量化）
Gregg, thanks for the question on that. Yes, on the cloud side, again, we saw strong performance reaccelerating to 29% year-on-year there. And maybe I'll start with the fact that DC migrations again to the cloud were in line with our expectations and not the realistic contributor to that $50 million beat on the print. It's progressing well, and we still expect that to contribute that mid- to single high-digit growth to cloud, like we mentioned. So the real 2 primary drivers that we talked about on the outperformance is really that cross-sell and seat expansion. . So on the cross-sell side, we saw outperformance in our collections business across Service Collection and in particular, Teamwork Collection. We have got Jira and Confluence, Loom and Rovo. And just keep in mind that right now, Teamwork Collection is really the best vehicle for our customers to buy and unlock AI and all the agent capabilities across the Atlassian platform, right? Customers are upgrading to TWC because of the increased AI credits. We're giving 10x more credits on Rovo versus the stand-alone subscription. And I'm sure Mike can touch a little bit more on some of the progress that we're seeing there. But importantly, we're also -- the other driver here is that we're seeing that growth in TWC while also seeing continued seat expansion in our core Jira stand-alone offering, right? And I think that really speaks to how an AI-driven agentic world, Jira and the Atlassian platform really remain core to enabling customers to manage their workflows and collaboration to fully unlock that value of AI. So whether it's TWC or standalone Jira offerings, we're launching a ton of new value that's really enabling our customers to deploy agents to do the work, to capture that agent activity alongside work history, full permissions and audit trails and admin governance. So a lot of great traction here, as you can see in our Q3 print.
- 反对 正在计划迁移的 DC 客户 seat 扩张放缓，预计 DC 扩张更弱
So as a result, that drove greater-than-expected upfront term license revenue recognized in the quarter. In fact, relative to our expectations for Q3, we recognized approximately $50 million more in upfront term license revenue. And that's some of the trends that we've been seeing since our announcement of end of life back in September, but more pronounced in Q3 given the size of the expiry base and the pricing catalysts that I mentioned. And maybe lastly, the cohort of DC customers that are actively planning and transitioning to the cloud, we're seeing these customers moderate their seat expansion. versus historical trends we've seen. Again, I'll mention that retention rate remains incredibly high, but now expecting a more muted level of data center expansion from these customers going forward as they try to move to the cloud. We're still seeing really nice uplift when customers move from DC to cloud. So net-net, what we're seeing is that our largest strategic customers continue to deepen their commitment at lasting, whether that's on DC or cloud, and we're working hard to meet them where they are and help them accelerate that transition so they can unlock all the AI and agent capabilities in the cloud.
- 支持 管理层称未见 seat 压缩，反而是强劲扩张，NRR>120% 且环比上升
- Mgmt framed the upside as seat expansion plus cross-sell, not a one-off migration beat.- Rovo added millions of MAUs, AI Rovo credit usage grew >20% m/m, Rovo customers grew ARR ~2x non-Rovo customers, and Teamwork Collection is becoming the main AI monetization vehicle through larger embedded credit pools and more agent usage. The print weakens the “AI kills seats” thesis because Jira seat expansion and TWC upgrades both contributed to Cloud outperformance- Mgmt said it is seeing “the opposite” of seat compression, with strong expansion, strong cross-sell, strong AI usage, and NRR north of 120% with another sequential uptick.- Service Collection surpassed $1B ARR, grew >30% y/y, and mgmt called this TEAM’s largest-ever quarter for competitive displacements from a major ITSM provider (neg read-through to NOW)- Data Center Revenue $560.7M, +44.3% y/y vs Street $517.1M, +33.1%, benefited from ~$50M of greater-than-expected upfront term license revenue, helped by EOL dynamics, the large Q3 expiry base, and a March pricing change. Mgmt said Cloud migrations are on track and retention is robust, but Data Center customers preparing to migrate are moderating seat expansion, making FY27 DC deceleration steeper.- Non-GAAP OM was 34.0% vs Street 27.7%, and FY26 OM guide was raised to ~29.0% vs prior ~25.5% and Street ~25.6%. Bulls view this as evidence of Cloud platform leverage, restructuring savings, and expense discipline. Bears note the $50M upfront DC revenue benefit naturally lifts margin and Q4’s restructuring savings should not simply be extrapolated into FY27.- FCF $561M missed Street, with the miss tied to restructuring payments and billing/collection changes, including Data Center contract duration moving toward one year and annual billing on some multi-year Cloud contracts. Billings were also soft versus Street, but mgmt is trying to steer investors toward subscription ARR disclosures at the upcoming investor day to normalize revenue-recognition noise.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 口径不齐 准备迁移的 DC 客户在放缓 seat 扩张，FY27 DC 减速更陡
- Mgmt framed the upside as seat expansion plus cross-sell, not a one-off migration beat.- Rovo added millions of MAUs, AI Rovo credit usage grew >20% m/m, Rovo customers grew ARR ~2x non-Rovo customers, and Teamwork Collection is becoming the main AI monetization vehicle through larger embedded credit pools and more agent usage. The print weakens the “AI kills seats” thesis because Jira seat expansion and TWC upgrades both contributed to Cloud outperformance- Mgmt said it is seeing “the opposite” of seat compression, with strong expansion, strong cross-sell, strong AI usage, and NRR north of 120% with another sequential uptick.- Service Collection surpassed $1B ARR, grew >30% y/y, and mgmt called this TEAM’s largest-ever quarter for competitive displacements from a major ITSM provider (neg read-through to NOW)- Data Center Revenue $560.7M, +44.3% y/y vs Street $517.1M, +33.1%, benefited from ~$50M of greater-than-expected upfront term license revenue, helped by EOL dynamics, the large Q3 expiry base, and a March pricing change. Mgmt said Cloud migrations are on track and retention is robust, but Data Center customers preparing to migrate are moderating seat expansion, making FY27 DC deceleration steeper.- Non-GAAP OM was 34.0% vs Street 27.7%, and FY26 OM guide was raised to ~29.0% vs prior ~25.5% and Street ~25.6%. Bulls view this as evidence of Cloud platform leverage, restructuring savings, and expense discipline. Bears note the $50M upfront DC revenue benefit naturally lifts margin and Q4’s restructuring savings should not simply be extrapolated into FY27.- FCF $561M missed Street, with the miss tied to restructuring payments and billing/collection changes, including Data Center contract duration moving toward one year and annual billing on some multi-year Cloud contracts. Billings were also soft versus Street, but mgmt is trying to steer investors toward subscription ARR disclosures at the upcoming investor day to normalize revenue-recognition noise.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 支持 剔除迁移后的 cloud seats 每季度、每年持续增长，管理层称曲线未放缓——但只给图、没给数字或增速
Customers aren't just buying in, they keep expanding. So let me give you an interesting chart. I think it will be interesting for you. Let's look at total cloud seats, excluding all migrations. No migrations involved. And despite the, oh, seat-based, it's going to disappear. We can get out your rulers if you want, I know you do. You can see visually, it's not slowing down. It's a pretty amazing curve. Again, this is excluding all migrations, cloud seats. They continue to grow every quarter, every year, durable growth as we talk about.
- 支持 cloud seats（不含迁移）每个季度、每年都在增长，管理层认为会继续加速
You can see visually, it's not slowing down. It's a pretty amazing curve. Again, this is excluding all migrations, cloud seats. They continue to grow every quarter, every year, durable growth as we talk about. Why? Because workflows continue to get more complex, more intertwined, Teams and companies are buying more into the Atlassian platform and getting exponential value as they increase their commitments. And that's really important, and we think that only keeps accelerating from here. The teams start orchestrating the AI agents alongside their people. You heard me talk a little bit this morning about the Teamwork Graph. I think we have more ahead of us that we can continue to do because of our investments in that platform.
- 支持 CFO：Q3 TWC 超预期的同时，单品 Jira seats 也在扩张（无数字）
And there's a couple of reasons for that. And the first is that we are incredibly well diversified, not just across tech teams, but non-tech teams, business teams. It's allowing us to play in these massive expansive TAMs with line of sight to $140 billion of opportunity. And you saw some of that momentum play out in Q3 with the outperformance that we saw in TWC while also expanding our Jira seats on our stand-alone offerings. You saw service collection grow to over $1 billion in ARR, growing 30% year-over-year. And that value, that ROI that we're delivering to our customers right now is positioning us really well across all of these segments. The second is how well Atlassian is positioned within that enterprise workflow stack, right, serving not just developers, but HR teams, finance teams, legal teams to help them collaborate and manage an ever-increasing amount of workflows.
- 支持 CRO：客户在核心产品上扩张 seats 并采用更多产品（未量化）
The good news is that they're not sitting still with us and neither is my team of salespeople. They're driving customers to expand in the cloud and expand across the Atlassian portfolio. Our team is focused on uncovering new opportunities for our customers and driving product adoption, also cross-selling into new collections and at the same time, closing complex high product deals as well. Now our customers, it's fair to say, have more options than ever before, and they are clearly voting with their wallets, and they are expanding seats across our core products and adopting our additional offerings. In terms of when we look at our collections, that certainly is being led by teamwork collection. And as Mike said, that is our primary motion for AI monetization and then being led by service collection, which you saw the numbers there in terms of the size of that business, and we're continuing to make progress there. And if you look at the slide behind me, you can see the outcome of the strategy.
- 支持 Q3 TWC 和 Jira 独立 seats 均超预期，开发者和业务 seats 都在增长（未量化）
Yes. Good question. I think I heard maybe 2 questions in that, first around sort of pricing positioning, pricing power around that. Historically, Atlassian has been very transparent with regards to our pricing. And we've always felt really good about the ROI that we drive for our customers. We consistently hear that from our customer base. And I think that's partly what's driven the ability to take share over time and be able to expand across the enterprises as well. It's one of the #1 things that we hear when it comes to the displacements that we've seen over the last several years here as well. One thing that -- I think, again, that I'm going to really appreciate is the ability for Atlassian to take the feedback from customers in terms of product, product parity, features, et cetera, be able to iterate on that quickly and still have really compelling price points to make sure that we're driving that adoption across enterprises and customers. So feel really good about the position that we have across that. And if you think about, again, the -- 85% of the Fortune 500 are customers of Atlassian, but that's only 10% of our revenue. It gives us a ton of room to continue to grow. The other area, maybe to your question here around like how that evolves over time. As it relates to commercial models, right, whether it is a per seat model, whether it's a consumption-based model, I don't think it's one size fits all. I mentioned earlier that Teamwork Collection is one of the best ways that our customers want to buy. And part of the reason of that is because customers are looking for some stability, right? As I'm sure you all have heard the term token maxing out there right now. And one of the things that we hear from our customers is, listen, we want some predictability in our billing structure, right? So we want to be able to buy in and unlock AI. And so they're doing that through Teamwork Collection because they're getting 10x more credits right now. But over time, as that consumption continues to grow, as those seats also continue to grow as we saw in Q3, right, Teamwork Collection outperformed as well as Jira stand-alone seats outperformed, and that was across both developer and engineering seats as well as the business seats as well, we'll continue to evolve the business models and the commercial models to meet those needs.
- 支持 3Q 强劲由 Jira seat 增长和 TWC 升级驱动
Wells Fargo highlighted TEAM as one of its preferred 2H software ideas, citing improving seat growth, accelerating Teamwork Collection adoption, and growing AI-related developer activity. The firm noted 3Q strength was driven by Jira seat growth and Teamwork Collection upgrades, with Teamwork Collection becoming a larger ARPU uplift than initially expected. Wells also sees incremental upside from datacenter migrations, Rovo adoption, and stronger developer hiring trends, arguing investor concerns around AI have largely shifted from being a headwind to a potential growth driver.
来源背景：Wells Fargo 卖方（TMTB 转述） - 支持 Q3 cloud 超预期来自 TWC 交叉销售和核心 Jira seat 扩张；剔除迁移后 seat 增长持续复利（图表，本稿无数字）
Yes. So we had a really great quarter. As Gregg mentioned, I think importantly, cloud revenue showed a lot of strength this quarter. The driver of that cloud outperformance was driven by 2 primary factors. One was cross-sell into Teamwork Collection as well as the second factor being seat expansion within core Jira. So those are great things to see, especially, I think, in the face of people having doubts about the durability of seat expansion. We actually recently showed a chart about seat expansion at our user -- or investor forum at our user conference, Team '26 in Anaheim. We showed kind of that seat expansion or seat growth ex migrations continues to grow, continues to compound over time. But -- and that really played out in Q3, right? So seeing that seat expansion across both developers as well as knowledge workers drove that outperformance. And then going back to Teamwork Collection, the other primary driver, the #1 reason why customers are upgraded to Teamwork Collection where you get the whole Atlassian platform as well as 10x the amount of Rovo credits with that. So you get a certain amount of Rovo credits with the paid Jira subscription or any standard subscription. But when you upgrade to Teamwork Collection, you get 10x the amount of credits. And that's really resonating with customers as they value that predictability as they kind of go forward and see their increase of Rovo usage.
- 支持 云超预期的两大驱动是 cross-sell 和 seat expansion；迁移符合预期，不是超预期来源
Thanks, Ryan. No, the philosophy remains -- really remains the same here. And on the cloud revenue drivers here on the $49 million beat, maybe I'll start with the fact that and highlight the fact that the DC migrations to cloud were really in line with expectations and not a contributor to the $49 million beat. It's progressing well and in line with our expectations for FY '26 contributing approximately mid- to high single digits to the cloud growth revenue. So the 2 primary drivers here of the outperformance were really the cross-sell and seat expansion that we were seeing similar to what we saw in the cross sell side, we continue to see strong uptake in our collections across Service Collection and Teamwork Collection. Service Collection revenue growth accelerated in Q4, just a strong signal of the AI adoption that we're seeing and that Mike mentioned a little bit earlier here. In fact, we're seeing over 3x increase in Agentic automation in Service collection as customers are continuing to deploy more agents beyond IT teams in HR and legal and finance and marketing teams. So again, also great to see the recognition here being named a leader in Gartner's Magic Quadrant for conversational AI platforms for employee services. We also continue to see a lot of strong performance on the Teamwork Collection side. A good reminder that Teamwork collection is still the best vehicle for customers to adopt and unleash AI on the Atlassian platform. This is where customers are upgrading to increased Rovo credits, it's giving customers what they want in terms of the predictability as well as the ability to really load balance those tokens across their organizations, and we're seeing that drive our strong ARPU uplift here as well. So really, I think what this shows is that the customers are consolidating their collaboration stack onto the Atlassian platform, recognizing the value that the Teamwork Graph really brings here. And we do see that in the results with Teamwork customers having 4 to 5x more paid seats per instance than stand-alone Jira or Confluence customers.
- 支持 TWC 增长的同时，Jira 和 Confluence 核心产品 seat 扩张持续强劲，且广泛分布
Thanks, Gregg. I can start with that and maybe throw it over to Mike. To Ryan's question a little bit earlier about some of the cloud outperformance drivers that we saw. I talked about the cross-sell side of it, and we are seeing customers continue to go wall to wall with Atlassian here and continue to adopt higher additions and move towards Teamwork Collection, Service Collection. But I think importantly, we're seeing that growth in Teamwork Collection while also seeing continued strong seat expansion in our core Jira and Confluence offerings here. And this expansion is broad-based. Recall that roughly 2/3 of users across Jira and Confluence are knowledge workers, right? So these are non-software development teams, really sitting in HR, legal, finance and really speaks to that diversified user base that we're seeing here. And I think this growth does demonstrate that in AI-driven world, the need for collaboration at the enterprise scale, tracking, planning, managing work, it's actually increasing. And Atlassian is a mission-critical platform here, helping our customers orchestrate their teams, their agents, their workflows to really unlock that value in AI and driving real ROI and real outcomes here. And that really was the other primary driver of the outperformance that we saw in cloud, including the cross-sell that I talked about a little bit earlier.
- 支持 CEO 称看到更强的 seat 扩张（相对口径未说明）
When we think about what AI is doing for Atlassian, the results speak for themselves. Customers are getting real value from the AI embedded directly into their workflows, and it's translating into direct growth in our business. Customers are committing more deeply to the Atlassian platform because of the value delivered. We're seeing longer customer commitments, larger deals, stronger seat expansion, great cross-sell momentum and higher NRR. Last quarter, we recognized customers like NVIDIA and ResMed at Team '26 for how they're putting the Atlassian platform to work in the AI era to drive their businesses using the Teamwork Graph. The reason customers like these and many more are getting so much value from Atlassian comes down to a key structural advantage and that advantage is context. We believe models will keep improving and organizations will hire that intelligence by the token.
- 支持 DC 客户迁移前 seat 扩张平淡（FY27 DC -17% 原因之一）
FY27 non-GAAP OM of 25% looks weak against Street at 28.8%, but includes roughly four points from lapping FY26 Data Center revenue-recognition benefits and another three-point headwind from moving compensation from SBC toward cash. Adjusting for both, underlying non-GAAP OM is expected to expand roughly 150-160 bps. * **Data Center** Data Center revenue of $462M, +21% y/y, beat Street by $47M due to greater upfront term-license recognition, pricing and strong retention following the EOL announcement. FY27 Data Center revenue is expected to fall 17% as TEAM laps that accounting benefit, FY26 purchasing pull-forwards and muted seat expansion among customers preparing to migrate. This will materially obscure the underlying Cloud trajectory during FY27. * **FCF** FCF was the clearest blemish at $475M, +32% y/y and a 26.9% margin, versus Street around $661M. The shortfall reflected severance, working-capital and billing-timing effects, including the move toward annual Cloud billing and more back-end-weighted enterprise deals.
来源背景：TMTB 作者转述业绩会与 callback - 支持 Q4 Jira 和 Confluence 核心产品 seat 扩张持续强劲（未量化）
Yes. Thanks for having me, Jason. Yes, we're really pleased with the strong Q4 results to end our fiscal '26. I think what you're seeing is customers really value and understand the value of the overall Atlassian platform with the Teamwork Graph, which is a living contextual layer underlying the platform as well as our overall system of work, basically living system of record and system of work to actually drive those workflows for organizations. To Jason's point, we had a really strong quarter on the cloud side of things. And that outperformance was really driven by 2 things. It's actually consistent with what we saw in Q3. So it's good to see that consistency carry through. But it's driven by strong upgrades and cross-sell to our Teamwork Collection, which is basically customers being able to purchase the entire Atlassian platform and primarily for additional Rovo credits. We get 10x the amount of Rovo credits with the Teamwork Collection. So customers are upgrading for that additional AI capability as well as cross-sell motion into our Service Collection. So that was really great to see. All the while, we're starting to see -- continue to see strong seat expansion across our core products of Jira and Confluence, which I think highlights the importance of collaboration and teamwork and coordination in this AI era, right? Like, we've talked about for quite some time. In the AI era, the need to track, manage, plan all your work across the organization that doesn't change. And I think you're starting to see that continue to play out with strong cross-sell momentum, AI purchasing on Teamwork Collection with that strong seat expansion on our core products with Jira and Confluence. And I think it's notable that the strong seat expansion was across both software development and knowledge workers. So at our investor forum, we recently shared about 2/3 of users on Jira and over 2/3, more like 70% on Confluence are knowledge workers and non-software developers. That trend continued across this quarter where that net new seat expansion happened across both those sectors. So that continues to be really healthy across both segments.
- 支持 Q4 超预期来自 Jira、Confluence、JSM 的独立 seat 增长（转录写成 'See growth'）
We posted 1.8 billion in revenue growth, 28% year over year. We posted ARR growth of 23% year over year with RPO growing 44% year on year. So again, you know this we got a question around, you know, what changed overnight, what changed in the quarter ? And the reality is it was really a lot of consistent a lot of consistent innovation and execution across our product teams, our engineering teams, our go to market teams to really deliver and drive those types of results and to the areas that really outperformed our expectations was standalone. See growth in our core products like JIRA, Confluence and JSM. We saw revenue we accelerate in our service collection in the quarter as well. So a lot of opportunities still ahead of us here. And so in terms of highlights, you know, I shared a couple of these stats earlier here, but in terms of being a driver for us, you know, the growth that we continue to see across our team, we're collection.
- 支持 前几季超预期由各应用独立 seat 增长驱动（转录 'suite growth'），回应 collection 蚕食担忧
Yeah. So really pleased with the momentum that we continue to see on the cloud side of the world. And you know, some of our biggest drivers remains expansion as well as that cross-sell motion that we talked about. So I know there is some concern about cannibalization as it relates to our collection business. But, you know, as we saw in the last couple of quarters, the outperformance was driven by standalone suite growth that we saw in our applications as well as continued growth into Brooklyn collection. It's still relatively early. We only introduced that over the last year or so. I believe that's going to continue to compound.
- 反对 未来 1-2 年开发者 headcount 只增 0-5%，纯按人头 seat 的开发工具将承压
核心不在于有多少开发者，而在于有多少基础设施和工作流任务需要自动化。如果对公司做一个方向性的量化，我认为未来一到两年内，开发者员工人数（headcount）的年增长率可能只有0%到5%，但软件产出和变更量可能会增长15%到30%。因此，传统的纯按人头席位（per-seat）收费的开发者工具将面临压力。但那些能够将AI Agent、自动化、工作流规模、安全治理以及基于用量（consumption-based）进行商业化变现的平台，反而有机会推动单开发者支出（spend per developer）上升。因此，真正的变化并不是AI导致开发者变少、进而引发开发工具支出下降；而是AI带来单开发者产出提高和更多机器生成的工作，促使开发工具定价逐渐从单纯的人类席位转向席位加Agent及用量收费。对于Jira、Confluence、CI/CD自动化和DevSecOps来说，这一趋势至关重要。
来源背景：Strategy Executive，美国头部电信运营商（前任职） - 反对 SMB 客户裁员导致 seat 下降约 20%（前 Atlassian 销售负责人）
Now they've gone from 50 to 40 people because they laid off a bunch of folks. All of a sudden, there's a decline of roughly 20%.
来源背景：前 Atlassian Global Head of Enterprise Sales（In Practise 推文引述） - 支持 CEO 称 Jira 与 TWC 刚经历最好的增长年，用户规模大幅加速
* Cannon-Brookes says Jira and the Teamwork Collection just had their best year of growth, accelerating at massive scale in users. * Atlassian sells products, including via DX, showing customers token spend by model and feature and where to cut it; he says enterprises are increasingly price-sensitive as AI becomes a growing budget line. * In Amsterdam, Atlassian announced agent-to-agent integrations with Figma and Databricks, so Confluence users can query Databricks or Snowflake through the Teamwork Graph, making Atlassian a hub rather than a bypassed destination (our read). * Atlassian runs many fine-tuned open-weight models, including a lot of NVIDIA Nemotron, and his biggest change of mind this year is inference prices falling further than expected, which suggests margin upside on AI features (our read).
来源背景：Atlassian CEO Mike Cannon-Brookes（摘要转述）
谁在加 seat：非开发者
判断：方向成立，非开发者确实占一半以上；但"增长最快的是非开发者"从来没给数，占比数字前后口径不同，不能拿来算趋势。
同一个指标，一年里说法变了四次：2025 年 6 月是"Jira 用户 52% 技术、48% 非技术"，第二天变成"50% 的 seats"，2026 年 5 月变成"Jira 约 2/3、Confluence 7/10、JSM 超 3/4"，6 月又变成"平台 2/3 的活跃用户"。如果 48% → 65% 是真实变化，在 Jira 用户整体增长 10% 的假设下，开发者用户要一年少 26%，这和管理层说开发者 seat 也在涨矛盾 (见 Information Add-on)。更可能是口径从"用户"换成了"活跃用户"或换了产品范围。
Service Collection 也一样：2 月说"超过 2/3 客户用于非 IT"，4 月说"60%"。外部唯一相关的一手案例是反例：Flock Safety 打算把财务、HR、采购的服务台给 AI 原生的 Console，而不是 JSM。
Yes. I think that's a great point to bring up. I think oftentimes people think of us as only serving software teams, and we do serve software teams, right? They're important persona into managing a certain type of complex workflow. But it's important to think about how it's not just developers that we serve. We've served -- or we shared data previously around, hey, of Jira users, what does that mix look like? What does that composition of users look like? And it's about 52% of those users are technical in nature. So those are developers, engineers or IT. And then 48% are nontechnical or knowledge workers. And I think that shows you a couple of different things. I think it shows that increasingly, you have line of business people that have to interface with their more technical counterparts as software becomes a bigger part of company strategies, and two, our ability to kind of spread horizontally across an organization after landing with a more technically oriented team. And so we've been able to kind of traverse both those and kind of grow consistently across both those vectors as we're able to serve more and more of those nontechnical teams. But I think our kind of domain expertise with how software gets built, knowing how teams work together to build technologies and it's important kind of domain expertise. And then that's starting to cross into IT, right? The lines between software development and IT are starting to blur more and more as IT is much more about managing different software and services as opposed to the hardware, different assets that you have across an organization.
Yes. And again, I think it's important to remember that fundamentally Atlassian is a collaboration company, right? Like all of our products facilitate collaboration. And so as the output increases of software because no doubt it will, there become all these new problems that humans have to collaborate together on and solve and Brian echoed all of that, right? Like how do you manage all this new software? How do you sell it? How do you get feedback from what customers want so that you can iterate on that software? Again, there's an abundance of new problems that get created. I think it's also important to remember that already today, 50% of our seats are technical in nature, so that's developers, engineering, IT. And already, we serve 50% that are nontechnical. So those business users, and we'll continue to push more in that direction. When you actually look at some of these SMB tech start-ups that you would expect to be much more kind of embracing or forward tech embracing, when you look at their patterns of seat expansion, it's been consistent for the past year. So that's good. And then also when you contrast that against SMBs of similar size in different industries, it's very consistent. So you don't see any deviation in kind of seat expansion patterns. Continue to see very healthy seat growth, continue to see absolute seat growth. I think the other interesting anecdote is when we look at some of these CEOs or people that make kind of bombastic proclamations, when you look at the underlying customer data, we actually see seat expansion within a lot of these companies. And so it's -- I think we feel really good as we continue to see seat growth because of all these initiatives and projects that continue to develop, as Brian said, but also as we kind of continue to push more to serving these nontechnical users.
It's for operations teams and service teams. When we talk about assets and buildings and laptops and everything else, we're talking about far more than developers. We're powering hundreds of millions of workflows every month. Now you talk us -- heard us talk before about how half our users are not in technology teams. And we've got a half technology, half nonbusiness. We shared that before. So how about this one instead? Break it down and let's look at developers versus everybody else, knowledge workers. In Confluence, 7 out of 10 users for that $1.5 billion-plus business that's growing very well in the cloud, 7 out of 10 of our users are knowledge workers, non-developers. In Jira Service Management, it's over 3/4. Even on Jira, it's roughly 2/3 of the users are non-developers. That's HR and finance and legal and marketing and operations staff and design. We spent two decades connecting these business teams with their technology teams, and we have a lot more to do. Again, a different way of showing that we are a well-diversified, durable growth business.
Yes. I think -- so we announced the end of life of our data center or on-premise platform, which will end of life in March of 2029. But even at that, I think we feel really good about being able to serve those customers with our cloud platform. We obviously have built an enterprise-grade unified cloud platform, as I mentioned earlier. We feel good that a lot of that is behind us, and we're now better able to serve these enterprises that are on our data center deployment. There's still some work that we have to do between now and March of '29, which is great because there's that long runway. And it's about having that open dialogue with customers that may not be able to move now and understanding our road map between now and then so that are we -- can we deliver what you need between now and then. When we think about migrations, we've obviously had a lot of migration success, and we've learned from the end of life of our server product several years ago. And we continue to see that those migrations tick up in the more recent years. More importantly, you're seeing those data center customers when they migrate to cloud, they're migrating to our cloud premium and cloud enterprise SKUs. So 93% of those data center customers, when they migrate, they migrate to the higher-level editions. And then actually, what's more important, once they land in cloud, we have significantly more products to cross-sell them to. We have higher-level additions, which we can kind of upgrade them to over time or upsell them to over time. And then it's significantly easier to add users in the cloud, right? I don't have to ping my IT admin and say, "Hey, can you add Gregg into my instance?" I can actually with little friction, click and add, Gregg immediately, bring in more users. And again, you get the benefit of the whole platform, underlying teamwork graph. So you get this compounding effect. And that's why I think you see such strong user growth continue on the Atlassian platform in spite of, I think, investor concerns. And we shared more recently at our Investor Forum another cut. Again, we serve all teams on the Atlassian platform. I think sometimes we have this misconception that we're just a developer company. And we showed 2/3 of the active users on our platform are knowledge workers, nondevelopers, non-engineers. And so that's an important aspect that we have the ability to serve increasingly nontechnical teams, these knowledge workers, certainly, that have to work in conjunction with those software developers that kind of are our heritage. But increasingly, that's a big part of it. And I feel that's why we have a lot of confidence earlier to your question on how -- why do you have so much confidence around kind of durability of things like user growth, cross-sell because, first of all, it's really early on, and we feel it's early on in our opportunity. And we're seeing these proof points as we grow into those nontechnical users over time.
How do we know that in terms of those knowledge workers? Well, if you look at the 1 billion knowledge workers on the planet, about 900 million of them are outside of technology, about 70 million technical knowledge workers and about 30 million of them are in development. Call these order of magnitude, rough numbers, there's various surveys they all end up about this sort of proportionality. If you think about the fastest-growing segment of our customer base, which is that nondeveloped audience, we have an awful long way to go with our tens of millions of users today to that $1 billion. So a few different ways you can think about our potential. And again, when we land a new department, we don't add a new customer. Just 2, 3 months ago, we landed a 200-person marketing organization that had never touched Jira in one of the 10 largest companies on the planet.
There's a couple of stages to that. We still do have Jira as our main IT support help desk for the IT support team. Console sits in between Slack and our help desk. If somebody submits it via email or directly through our portal, it just goes directly into Jira that doesn't get touched by Console. We do have some automations that we can use to basically redirect them to Console to use Console first, if it's a request or a ticket that we know Console can handle. There are plans to potentially move away from that model and move directly into Console as our one-stop shop. Not only for IT, we actually have plans potentially to introduce this particular tool to other support teams within our company. A lot of our team members have internal customers and Console really excels at that, answering questions, directing them to the right place and then also submitting tickets on behalf of users. This year we're actually looking to expand that into other teams that can include like finance, procurement, onboarding and offboarding HR, even our security team, things like that. That way, somebody goes to a single source and says, "Hey, I have this question or this request or this whatever," and they'll be directed to the proper authority for that particular request based on the build out that we do for each of those different teams.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户
不过，我不确定 Atlassian 能在多大程度上兑现全部承诺。从某种意义上说，我会将其与 Palantir 进行比较。Palantir 目前高度专注于企业市场，原因在于它此前曾在企业市场遭遇失败。Palantir 在政府市场表现非常出色，但这也让许多人产生疑问，不确定其产品是否真正具备所宣称的能力。我认为，Atlassian 目前正处于类似的关键阶段。
来源背景：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团；自称并非 Atlassian 专家
展开全部 39 条原文 (按时间)
- 支持 TWC 用来触达目前尚未覆盖的知识工作者（愿景）
I just wanted to add, Keith, from my point of view, maybe slightly more philosophical than Joe. Look, collections are about being as customer first as we can, and we always try to do that at Atlassian. It simplifies how our customers can buy and grow with our offerings. That simplicity, that reduction in friction does result in, in our experience, a long-term growth driver that appears over time as customers increasingly want to buy larger amounts of software from Atlassian, that they want to not buy it in smaller and smaller pieces, if that makes any sense. They want to buy it in large amounts. That enables us to create quite some significant simplicity in the purchasing behavior, but also simplicity in the deployment and the rollout behavior. As Joe mentioned, that simplicity comes from having a single addition, having a single seat count, and any new offerings that are introduced like Rovo, for example, into the Teamwork Collection will just appear for those users, which is a much easier experience for the customer. I believe, over time, that will lead to continued growth, that's part of the reason, but also lead to customer stickiness and customer happiness as more of our functions are exposed to more people within that organization. And it gives us a larger opportunity to go after the knowledge workers that we don't have access to in many of our customers at the moment or who don't have access to our products. So philosophically, I think it's about simplifying things for our customers in terms of their purchasing behavior.
- 支持 未来用户不再是原来那批人，可服务的人群和类型在扩大，并称已经看到证据
Yes. This goes back to the point around what happens in the future? So what do you believe AI will help people do? Fundamentally, everything we've seen so far, all the evidence points to a place where more things get built, more content gets created, more software gets created, more people are able to create things than we were able to do before. Now Atlassian is in the business of teamwork. So how do you help people create these new things? How do you help people create new software? Is it going to be the same people as we used to serve before? No, and that's a good thing because it's expanding the number of people we can serve, the kind people we can serve, and we see that evidence already. So it shows up in the business today already. Jira Service Management is one of our fastest-growing business, over $600 million. And Jira Service Management gets used by a lot of different teams who have nothing to do with software development. These are HR teams, these are marketing teams. The teams that are actually super charged by AI, they use our products because it makes them better. And the licensing model, specifically, we have consumption-based pricing, and we've had this for quite a while. But I do think consumption-based pricing will get more interesting and important in an AI world because as people unlock more productivity and as people learn to do more things with AI, they will also consume more of these kinds of capabilities. So we already have 70 different kinds of consumption-based pricing levers built into the product in Atlassian into the product portfolio. And Rovo is, our AI offering, also comes with a bunch of consumption- based meters. And I think those things will get more interesting in terms of how does the business evolve. But fundamentally, as usage increases, the business is going to benefit from this overall.
- 支持 超过一半的使用量来自非开发团队
Sure. So Atlassian is a 23-year-old company, we serve about 300,000 customers, which is an unusually broad swap for companies that serve B2B, that build B2B software. And about 85% of Fortune 500 companies use our products, and we are very ubiquitous across teams of every kind. Atlassian's mission is to unleash the potential of everything. So in fact, funnily, when I joined Atlassian 10 years ago, I joined thinking, "Oh, this is about Jira, which is project management tool, the tracking tool, right, to your Bugzilla reference." And the first week, I was shocked when I looked at product analytics, that half -- well over half of our customers, in fact, had nothing to do with development teams or tracking or any of that. Jira was being used for tasks as varied as hiring people, tracking recruitment loops, tracking financial workflows, tracking how customer support was getting handled, tickets with internal health desks, so a lot of collaboration workflows. So fundamentally, we are a collaboration company. And it started out with developer teams. So the original set of teams that used to adopt our products would be development teams, people who build software, so engineers and engineering leads and then it spread to product managers, designers and surrounding teams. But even now across -- we have a portfolio of about 14 different products and cloud platform that supports all of these products. Well over half of our usage still comes from different kinds of teams, but not developer teams necessarily.
- 支持 Jira 用户约 52% 是技术用户，48% 是非技术用户或知识工作者
Yes. I think that's a great point to bring up. I think oftentimes people think of us as only serving software teams, and we do serve software teams, right? They're important persona into managing a certain type of complex workflow. But it's important to think about how it's not just developers that we serve. We've served -- or we shared data previously around, hey, of Jira users, what does that mix look like? What does that composition of users look like? And it's about 52% of those users are technical in nature. So those are developers, engineers or IT. And then 48% are nontechnical or knowledge workers. And I think that shows you a couple of different things. I think it shows that increasingly, you have line of business people that have to interface with their more technical counterparts as software becomes a bigger part of company strategies, and two, our ability to kind of spread horizontally across an organization after landing with a more technically oriented team. And so we've been able to kind of traverse both those and kind of grow consistently across both those vectors as we're able to serve more and more of those nontechnical teams. But I think our kind of domain expertise with how software gets built, knowing how teams work together to build technologies and it's important kind of domain expertise. And then that's starting to cross into IT, right? The lines between software development and IT are starting to blur more and more as IT is much more about managing different software and services as opposed to the hardware, different assets that you have across an organization.
- 支持 业务人员越来越多地要和技术人员协作，Atlassian 能从技术团队横向扩展到非技术团队
Yes. I think that's a great point to bring up. I think oftentimes people think of us as only serving software teams, and we do serve software teams, right? They're important persona into managing a certain type of complex workflow. But it's important to think about how it's not just developers that we serve. We've served -- or we shared data previously around, hey, of Jira users, what does that mix look like? What does that composition of users look like? And it's about 52% of those users are technical in nature. So those are developers, engineers or IT. And then 48% are nontechnical or knowledge workers. And I think that shows you a couple of different things. I think it shows that increasingly, you have line of business people that have to interface with their more technical counterparts as software becomes a bigger part of company strategies, and two, our ability to kind of spread horizontally across an organization after landing with a more technically oriented team. And so we've been able to kind of traverse both those and kind of grow consistently across both those vectors as we're able to serve more and more of those nontechnical teams. But I think our kind of domain expertise with how software gets built, knowing how teams work together to build technologies and it's important kind of domain expertise. And then that's starting to cross into IT, right? The lines between software development and IT are starting to blur more and more as IT is much more about managing different software and services as opposed to the hardware, different assets that you have across an organization.
- 支持 已有 50% seats 是技术类，50% 是非技术业务用户，还会继续往非技术方向推
Yes. And again, I think it's important to remember that fundamentally Atlassian is a collaboration company, right? Like all of our products facilitate collaboration. And so as the output increases of software because no doubt it will, there become all these new problems that humans have to collaborate together on and solve and Brian echoed all of that, right? Like how do you manage all this new software? How do you sell it? How do you get feedback from what customers want so that you can iterate on that software? Again, there's an abundance of new problems that get created. I think it's also important to remember that already today, 50% of our seats are technical in nature, so that's developers, engineering, IT. And already, we serve 50% that are nontechnical. So those business users, and we'll continue to push more in that direction. When you actually look at some of these SMB tech start-ups that you would expect to be much more kind of embracing or forward tech embracing, when you look at their patterns of seat expansion, it's been consistent for the past year. So that's good. And then also when you contrast that against SMBs of similar size in different industries, it's very consistent. So you don't see any deviation in kind of seat expansion patterns. Continue to see very healthy seat growth, continue to see absolute seat growth. I think the other interesting anecdote is when we look at some of these CEOs or people that make kind of bombastic proclamations, when you look at the underlying customer data, we actually see seat expansion within a lot of these companies. And so it's -- I think we feel really good as we continue to see seat growth because of all these initiatives and projects that continue to develop, as Brian said, but also as we kind of continue to push more to serving these nontechnical users.
- 支持 分析师观点：bundle 能触达更多 seats，扩大在组织内的覆盖
William Blair & Company L.L.C., Research Division Yes. So it gets you access to a lot more seats than I think is a thesis, right, that you can broaden your footprint inside these organizations. When you think about -- you have 330,000 customers. So when you think about the opportunity to actually target which customers are best for the bundles, for the Teamwork bundle to expand into the CFO, into the HR organization. How do you think about doing that? Is that a particular subset? Or is this something that every Atlassian customer -- it's relevant for every Atlassian customer?
- 支持 CEO 判断：财务、HR、市场等其他职能会有更多人写软件
Let me give you some assumptions that I have or some beliefs. Do I think there will be far less developers in the world 5 years from now? No, I don't think so. I think there'll be far more and far more people creating software in other functions, right, whether they're in finance or HR or marketing, there's going to be a lot more people creating software. Do I think developers roles change and augment with these tools? Yes. Develops as use of that.
- 支持 核心产品用户中 50% 是业务用户
This partnership marks a major milestone in Atlassian's multi-cloud strategy, accelerating our cloud transformation and delivering advanced AI solutions to millions more users worldwide. In the AI-era, we believe the need for collaboration increases significantly as more people are able to create and more ideas can be brought to life. This creates a huge opportunity for Atlassian. Our platform powers teamwork, business processes and workflows across all types of teams, but now 50% of users of our core apps being business users, reinforcing the mission-critical role that our platform plays and helping both businesses and technology teams collaborate. If you'd like to hear more about how this sets us up to win in the AI-era, check out the Loom that I just posted to our IR website. This quarter, we saw more enterprises realize the power of going wall to wall with the Atlassian system of work, with many of them consolidating from a variety of other tools onto the Atlassian Cloud platform. The result with continued revenue growth across our core apps of Jira, Confluence, Jira Service Management and Loom, all of which are growing in line or faster than total company revenue with significant runway ahead of all.
- 支持 业务用户板块几乎每季度都在增长，且在加速（未量化）
Secondly, we see continued progress in the business user segment. You're right at calling out there are many more business users out there. We see that as a huge opportunity for us. We've grown that almost every quarter for the last few years, but it's certainly accelerating. Why is that? Lots of things. So for example, the Teamwork Collection is a broader offering separating out our software pieces from our broad collaboration pieces really helps with that.
- 支持 技术团队使用量也在持续增长，业务用户增长不是以牺牲技术用户为代价
Whether that's in service management, whether that's in broad business workflows in Jira or whether that's coming Confluence and a wholesale of our applications together with all of these third-party tools. We just see that we have great momentum in continuing to attack that market. The business team usage continues to grow for Atlassian. I will say that the technology team usage always continues to grow as well. So it's not one at the expense of the other, which is really important because that's where you often get this proportionality problem right. Our software business is doing really well. And so it's not necessarily one for the other, but we're trying to do both simultaneously.
- 支持 增长最快的用户群在业务端（没给数字）
Yes. Great question. So firstly, a huge level of interest from our customers. I think the evolution that we've made as a go-to-market sales organization is we've really pivoted on the value conversation. which I would say, had been something that we've invested heavily in since I've arrived. Now we're focused on really -- Anu touched on it like 50% of our users are actually in the business, which for me coming into Atlassian was a real surprise. I'm sure it's a surprise for many people out there as well. And actually, what's a great position of strength to be in is that actually, the fastest-growing pool of our users is actually in the business as well. So where we're spending our time and our focus is on really understanding what is the value that our business users are deriving from the usage of our software. So now when we look at teamwork collection before we even start pitching system of work, it's to really understand what is...
- 支持 50% 的用户在业务端
Yes. Great question. So firstly, a huge level of interest from our customers. I think the evolution that we've made as a go-to-market sales organization is we've really pivoted on the value conversation. which I would say, had been something that we've invested heavily in since I've arrived. Now we're focused on really -- Anu touched on it like 50% of our users are actually in the business, which for me coming into Atlassian was a real surprise. I'm sure it's a surprise for many people out there as well. And actually, what's a great position of strength to be in is that actually, the fastest-growing pool of our users is actually in the business as well. So where we're spending our time and our focus is on really understanding what is the value that our business users are deriving from the usage of our software. So now when we look at teamwork collection before we even start pitching system of work, it's to really understand what is...
- 支持 AI 使用广泛覆盖业务团队和技术团队
All up over 300,000 customers, including Databricks, Expedia, Ford and Wells Fargo rely on Atlassian's AI-enabled cloud platform to power their business processes and mission-critical workflows. We're proud of our ability to continue to deliver AI into the hands of those customers to use today. We've amassed over 3.5 million monthly active users of our AI capabilities across the platform, once again, up over 50% since last quarter. This usage is widespread across both business teams as well as technical teams. I'll repeat what you've heard me say in the past, AI is one of the best things that's ever happened to Atlassian. They need to track, plan and manage work while harnessing your organizational knowledge are things that don't change in this era. And I'd argue those things become even more important as more software is created and more people have the ability to create amazing technology that changes our lives.
- 支持 超过 2/3 的 Service Collection 客户用于非 IT 场景
As such, in the last 6 months, more than 40% of the Agentic workflows that have been built are actually in Service Collection customers and Service Workflows. It's a very natural area to deploy agents -- AI agents into your service workflows to help improve the human agentic experience or the human agent experience of delivering value to the customer or the customer just getting the value directly themselves. So that is going very, very well. More than 2/3 of our service collection customers are using it for non-IT use cases at the moment, which is a great sign that, that is happening. Two other things maybe, one, you see we're a leader in our enterprise service management wave. The analyst community continues to recognize us as a leader and a visionary, but also a significant challenger and growth brand. And we continue to see a lot of migration from legacy service management platforms on to service collection for much higher ROI, much better cost equation with a much more modern stack and user experience, and that's really great for us.
- 支持 Service Collection 在 HR、财务等传统 IT/Ops 以外的团队增长
Ittai, if I can just chime in at sort of a nonfinancial and high level. Look, we're very clear that the system of work is about a continued growth of Atlassian into the knowledge worker population, right? When we talk about unleashing the potential of every team, our mission for over 2 decades, I would stress the every team part. We've got the software collection, which does very, very well. The DX and a lot of other things encompass and a bit like a pipeline, like there's a lot of great areas of that software collection. We've also got the Service Collection. So we're seeing great growth in Service Collection across HR teams, Finance teams, other areas outside of its traditional market in IT and Operations teams, but that's certainly an area of growth for us. So there's HR and Finance teams, we're seeing a lot in the Service Collection, for example, and you've seen us ship a lot of features. We also launched Customer Service this quarter as an application within that to go after a new set of teams we haven't been able to get to as well, you would say. In the Core Teamwork Collection, look, as we said in our letter, we were seeing double-digit seat expansion, more than double-digit seat expansion compared to people who buy the stand-alone applications. And that's baked into the Teamwork Collection packaging, but it's also because of the AI offering and how it works, right, in terms of getting -- you get -- you equate your Loom Confluence and Jira seats along with your additions, which often gains expansion. But given the nature of our applications and our continued growth in business teams across an Enterprise sales, Marketing, HR, Finance, et cetera, what that licensing structure allows those teams to do is to collaborate more. And collaboration is a very sticky and kind of viral activity. So that's where you're seeing that expansion coming through in our MAU and in our AI MAU and also ultimately ending up in our NRR and RPO numbers in terms of long-term commitments from customers that, that is where they're seeing expansion across business teams of all struts.
- 支持 一半用户在业务端，业务端是增长最快的部分
Yes. So you're right, 85% of the Fortune 500 are our customers, only represents 10% of our overall revenue. Pretty amazing that Atlassian did this without a massive go-to-market team. I always remind myself that every single one of our customers entered through the exact same door that is they swipe their credit card online and then they came in as a customer. And now my role is to build out our go-to-market team. By the end of this year, we will have approximately 400 sellers, which again is not a huge go-to-market team. That has pretty much doubled from where we were last year. The approach that we have is we need to continue to hire more capacity so that we can increase our coverage. We continue to revitalize the ecosystem so that we have them better aligned with our goals. We continue to focus on moving our customers to the cloud, which obviously is going to unlock more potential for us and moving our customers to higher addition values, which also is going to help us unlock more value. And to your case in point around the German automotive customer, one area that's a massive focus for us is moving from IT to the business. So exactly that German automotive customer of 400,000, they clearly don't have 400,000 developers because half of our users are actually in the business, and the business is actually our fastest-growing segment. I know many always think that we're only in dev and IT, but we are beyond dev and IT, and we're going to continue to focus on growing outside of dev and IT as we move forward.
- 支持 Jira 用户 50% 技术、50% 非技术；开发者只占业务的一小部分
Yes. I think it's interesting, right? We have a strong heritage with developers, and that's where we got our start initially. That represents a small part of our business today. We actually look at even Jira, Jira's core user base. 50% of their users are technical. That includes basically developers, engineers and IT. The other 50% are nontechnical or general business users. So we already have a very diversified kind of user base. Everything -- everyone is focused right now on the development and how that continues to evolve. We think of this world where humans and agents are going to work side-by-side. Humans are always going to have to be in the loop. When you think about humans are always going to have to have an agency of directing what do we want these agents to do. What do we want them to build? What do we want them to accomplish? And then humans will have to have checkpoints of, did that agent actually accomplish what we wanted it to do. So there continues to be this role of humans and agents in the loop, especially even in software development. And again, tracking back to the point I made earlier, the role that Jira plays, I don't think that changes anything that level of criticality continues to rise because you need to be able to say, okay, did that agent actually build what we wanted to do? How do I track and manage all these different agents running across my business? Are they doing what we wanted them to do. Jira has always been this orchestration layer of tracking work and who's working on, what is the output. Does that actually ladder up to our company strategy in a world where there are agents in the mix, are those agents doing that? How do I track this increased complexity across my organization. And so again, I think that's a really exciting opportunity for us.
- 反对 非 IT 部门（财务、采购、HR、安全）的服务台扩张计划给 Console 而非 JSM
There's a couple of stages to that. We still do have Jira as our main IT support help desk for the IT support team. Console sits in between Slack and our help desk. If somebody submits it via email or directly through our portal, it just goes directly into Jira that doesn't get touched by Console. We do have some automations that we can use to basically redirect them to Console to use Console first, if it's a request or a ticket that we know Console can handle. There are plans to potentially move away from that model and move directly into Console as our one-stop shop. Not only for IT, we actually have plans potentially to introduce this particular tool to other support teams within our company. A lot of our team members have internal customers and Console really excels at that, answering questions, directing them to the right place and then also submitting tickets on behalf of users. This year we're actually looking to expand that into other teams that can include like finance, procurement, onboarding and offboarding HR, even our security team, things like that. That way, somebody goes to a single source and says, "Hey, I have this question or this request or this whatever," and they'll be directed to the proper authority for that particular request based on the build out that we do for each of those different teams.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户 - 支持 Teamwork Graph 提供的共享上下文让 Atlassian 以越来越快的速度扩展到非技术岗位（未量化）
Yes, Gregg, thanks for that and for calling it out. I hope we've been very consistent on our views in that world. We are not seeing any signal of seat compression from customers. If anything, we are seeing the opposite. We are seeing strong expansion numbers, strong cross-sell numbers between collections, strong usage of AI and strong commitment to the Atlassian platform. Many, many competitive wins, a huge amount of consolidation happening into the Teamwork Collection. So we have a lot of green lights in a lot of different places. Similarly, you can see that NRR maintained north of 120% and even ticked up again for, I think, the third or fourth quarter in a row. We have a lot of reasons why that is. Firstly, I would go to high R&D investment in just great quality software, right? That does matter. It always gets ignored in a lot of context, wrong choice of words, in a lot of quarters. But we build amazing applications that deliver great value for our customers. Secondly, the context we have in the Teamwork Graph and the critical business processes continue with AI to blur the boundaries between teams and between roles. That means our platform and our offering between service teams on one side that connect to finance and HR, marketing and in through ops teams to technology teams with business teams and leadership teams. The same context across all of those teams allows us to expand into those nontechnical roles at an increasing rate, right, which shows why we are getting that seat expansion in different collections, in different areas and just the continues in our business. Fundamentally, customers are opting for more and more workflows on the Atlassian platform.
- 支持 60% 的 Service Collection 客户在 IT 以外（HR、市场等）使用
Thanks, Brent. Yes, we had a great quarter for competitive displacements, especially in the Service Collection, especially -- look, we continue to be incredibly strong in the mid-market area. As you can see from many years of investment in the enterprise pillars of our business, we are starting to go really, really strongly in the enterprise and strategic segments across service management in particular. That's not just in ITSM, although in ITSM, we are growing really strongly, it is in broader employee service management. We're seeing that -- we get the statistic of the 75% of the Fortune 500 uses Service Collection. 60% of Service Collection customers use us outside of IT on HR and marketing in other areas. This is just a fantastic example and why we're getting those competitive displacements. It was our largest quarter ever for those. And it's because, again, I would go to the quality of the software, the state with which you can get up and running on the Service Collection continues to be a strength. The high level of user experience quality continues to be a strength. The comprehensiveness of our data and the Teamwork Graph and bringing that to their Service Collection fundamentally allows you to operate those services cheaper, quicker, better answers because we have access to a better knowledge graph across your organization. And AI continues to win every which way. We're incredibly AI forward. It's one of the largest areas of usage of agents in automation and automated workflows because it allows the service teams to run more quickly. And we're only just getting started in customer service and had a great quarter in that area as well. And our asset management platform, again, assets moving into the platform as a whole, part of that overall Teamwork Graph. It's just an incredibly strong customer story. And so we're really excited about how we keep taking share in that space. And $1 billion of ARR is a great milestone. We thought it was worth celebrating. That's just a fantastic milestone for that business while continuing to grow incredibly fast.
- 支持 新增 seats 不是开发者驱动；增长早已由非开发者带动
You heard me talk a little bit this morning about the Teamwork Graph. I think we have more ahead of us that we can continue to do because of our investments in that platform. Brian? Now you might be asking, okay, they're adding seats, but seats for whom, right? And this is where the old narrative, I think, breaks down. People pigeonhole Atlassian as a developer company. I might argue there's going to be more developers in the AI era. I think that's going to be true. That's fine. You think we're a developer company, or some people do. We love that heritage, I will say. Developers are awesome. Technology and business teams working together is the core of what we do. We think developers are incredibly important to the future of every single large business on this planet. Technology is the core strategic advantage of almost every single big company that you will talk to. If not, they have some sort of regulatory barrier, et cetera. But the growth we're seeing hasn't been driven by developers for a long time. The system of work is for the entire company. It's for finance teams, it's for HR teams, it's for marketing teams. It's for operations teams and service teams.
- 支持 用户结构：Confluence 7/10、JSM 超 3/4、Jira 约 2/3 为非开发者；此前披露约一半用户不在技术团队
It's for operations teams and service teams. When we talk about assets and buildings and laptops and everything else, we're talking about far more than developers. We're powering hundreds of millions of workflows every month. Now you talk us -- heard us talk before about how half our users are not in technology teams. And we've got a half technology, half nonbusiness. We shared that before. So how about this one instead? Break it down and let's look at developers versus everybody else, knowledge workers. In Confluence, 7 out of 10 users for that $1.5 billion-plus business that's growing very well in the cloud, 7 out of 10 of our users are knowledge workers, non-developers. In Jira Service Management, it's over 3/4. Even on Jira, it's roughly 2/3 of the users are non-developers. That's HR and finance and legal and marketing and operations staff and design. We spent two decades connecting these business teams with their technology teams, and we have a lot more to do. Again, a different way of showing that we are a well-diversified, durable growth business.
- 支持 非开发者是增长最快的客户群；当前数千万用户 vs 约 9 亿非技术知识工作者的空间
How do we know that in terms of those knowledge workers? Well, if you look at the 1 billion knowledge workers on the planet, about 900 million of them are outside of technology, about 70 million technical knowledge workers and about 30 million of them are in development. Call these order of magnitude, rough numbers, there's various surveys they all end up about this sort of proportionality. If you think about the fastest-growing segment of our customer base, which is that nondeveloped audience, we have an awful long way to go with our tens of millions of users today to that $1 billion. So a few different ways you can think about our potential. And again, when we land a new department, we don't add a new customer. Just 2, 3 months ago, we landed a 200-person marketing organization that had never touched Jira in one of the 10 largest companies on the planet.
- 支持 新部门落地（例：全球前十大公司 200 人营销团队首次用 Jira）不计为新客户，体现为存量客户内 seat 扩张
Call these order of magnitude, rough numbers, there's various surveys they all end up about this sort of proportionality. If you think about the fastest-growing segment of our customer base, which is that nondeveloped audience, we have an awful long way to go with our tens of millions of users today to that $1 billion. So a few different ways you can think about our potential. And again, when we land a new department, we don't add a new customer. Just 2, 3 months ago, we landed a 200-person marketing organization that had never touched Jira in one of the 10 largest companies on the planet. What was important is those 10 largest companies, they said they had never used Jira. In fact, one of the people said, I didn't know this was Jira. I thought this was someone else.
- 支持 销售正以 persona-based selling 从 IT 扩展到市场、HR 等新买方
Yes. So obviously, we have considerable customer still that we need to move from DC to cloud, and we are continuing to make progress there. We have also established a considerable specialized sales organization. Now that specialized sales organization allows us to focus more in terms of selling into specific buyers into our customers. We historically have landed in IT, so tech. And from tech, we will expand into new buyers. So when we now look to sell to new buyers within the organization, we have to position ourselves differently to those particular personas. Therefore, we are selling into -- if it's marketing, if it's into HR, et cetera, we have persona-based selling that we have in place. So we are seeing and making progress there. Where from a collection perspective, obviously, Service Collection is where we are making considerable progress. We have -- as you saw in our Q3 results, we have made considerable progress in Service Collection. And then also, we have had seen a pickup in terms of our win rate against [ ServiceNow ], specifically and also some replacements as well in terms of our competitors, which is what we like to see. And we're hopefully expecting that that's going to continue in that vein.
- 支持 客户（Canal Plus CTO）称因 Rovo 决定将 Atlassian 套件推广到非技术部门
Yes, exactly. IT, in fact, at the beginning, we use like a lot of company confluence because we have all the documentation. And after we start to use the different suite of Atlassian and last year, with Team '25 when we discovered Rovo, we said, okay, this is a key now to be sure that we can deploy all the suite for all the departments, which are not technical.
- 支持 Jira 用户 65% 是非开发者知识工作者，Confluence 70%，JSM 75%
And I'll speak to some of those areas. I think the first is just how I underappreciated, I think, how well diversified the business really is. For those maybe not following Atlassian closely, you could sort of pigeonhole it into a technology point solution for engineers, right? And while there is an incredible legacy of product-led growth that really started in that domain, when I looked much more closely to the opportunity and where it's at now and frankly, where it's been for several years now, is that -- Jira users, as an example, 65% of Jira users are knowledge workers. These are non-edge, non-dev. These are folks and teams sitting across HR teams, finance teams, legal teams and marketing teams. And that number is even higher for Confluence at 70%. And for JSM, our Service Collection, that's 75%. So an incredibly strong and diversified base in terms of the opportunity set across enterprises and companies. And that diversification, I think, also shows up in other dimensions in terms of the scale of the types of customers that we have, right, over 350,000 customers, everyone from SMB, all the way to enterprises as well. The other aspect is when I think about fundamentally where AI is now and where it's going and the companies that are going to be incredibly well positioned to deliver that type of value for customers durably.
- 支持 Jira 用户 65% 非开发者，Confluence 70%，JSM 75%
Yes. Well, first, thanks for having me, Koji, and thanks, everyone, for attending today. When I was thinking about the next opportunity, really at the core of it was trying to understand which opportunity, which companies were really going to be a net beneficiary of AI. And when I looked across the board, Atlassian really jumped to the top of that list. And a couple of things really stood out for me. One was just how diversified and durable I felt the businesses were. And I think it's an area that I really sort of underappreciated coming into it because it's easy if you don't follow the company to think about it as the Jira company. It's a software tool only for developers and engineers. But what I soon discovered was that it was much more diversified than that. If you look at our Jira customer base right now, the usage is 65% nondevelopers. So these are nonengineers, nondevelopers. These are teams that are sitting in marketing teams, business teams, finance teams and HR, and that number is even higher for Confluence users at 70% and 75% for JSM, our Service Collection. And these are skilled businesses. This past quarter, we shared that JSM was a $1 billion ARR business, growing 30% year-on-year. Confluence, which no one really talks about, it's a $1.5 billion business, and Jira at $2.5 billion. So these are scaled businesses that are incredibly diverse and allows us to play in these expansive TAMs across multiple domains within the enterprise. That was an area that really stood out for me, Koji. And then I would say the other one is, if you think about fundamentally the types of companies that are going to benefit from AI and what's going to be an AI accelerant, it was going to be around context, right, which companies really have and harness context around enterprises and organizations. And when I think about the System of Work that Atlassian is developing and has sort of that center of gravity across different enterprises and where they sit in that tech stack across multiple, multiple teams, that context becomes incredibly important to be able to build the rails on top of which workflow sit, agents it. And so it just felt it was incredibly well positioned as part of that.
- 支持 seat 扩张覆盖开发者与知识工作者
Yes. So we had a really great quarter. As Gregg mentioned, I think importantly, cloud revenue showed a lot of strength this quarter. The driver of that cloud outperformance was driven by 2 primary factors. One was cross-sell into Teamwork Collection as well as the second factor being seat expansion within core Jira. So those are great things to see, especially, I think, in the face of people having doubts about the durability of seat expansion. We actually recently showed a chart about seat expansion at our user -- or investor forum at our user conference, Team '26 in Anaheim. We showed kind of that seat expansion or seat growth ex migrations continues to grow, continues to compound over time. But -- and that really played out in Q3, right? So seeing that seat expansion across both developers as well as knowledge workers drove that outperformance. And then going back to Teamwork Collection, the other primary driver, the #1 reason why customers are upgraded to Teamwork Collection where you get the whole Atlassian platform as well as 10x the amount of Rovo credits with that. So you get a certain amount of Rovo credits with the paid Jira subscription or any standard subscription. But when you upgrade to Teamwork Collection, you get 10x the amount of credits. And that's really resonating with customers as they value that predictability as they kind of go forward and see their increase of Rovo usage.
- 支持 平台 2/3 活跃用户是知识工作者；渗透非技术团队是增长重点
And we shared more recently at our Investor Forum another cut. Again, we serve all teams on the Atlassian platform. I think sometimes we have this misconception that we're just a developer company. And we showed 2/3 of the active users on our platform are knowledge workers, nondevelopers, non-engineers. And so that's an important aspect that we have the ability to serve increasingly nontechnical teams, these knowledge workers, certainly, that have to work in conjunction with those software developers that kind of are our heritage. But increasingly, that's a big part of it. And I feel that's why we have a lot of confidence earlier to your question on how -- why do you have so much confidence around kind of durability of things like user growth, cross-sell because, first of all, it's really early on, and we feel it's early on in our opportunity.
- 支持 Jira 和 Confluence 约 2/3 用户是知识工作者（HR、法务、财务等非软件开发团队）
Thanks, Gregg. I can start with that and maybe throw it over to Mike. To Ryan's question a little bit earlier about some of the cloud outperformance drivers that we saw. I talked about the cross-sell side of it, and we are seeing customers continue to go wall to wall with Atlassian here and continue to adopt higher additions and move towards Teamwork Collection, Service Collection. But I think importantly, we're seeing that growth in Teamwork Collection while also seeing continued strong seat expansion in our core Jira and Confluence offerings here. And this expansion is broad-based. Recall that roughly 2/3 of users across Jira and Confluence are knowledge workers, right? So these are non-software development teams, really sitting in HR, legal, finance and really speaks to that diversified user base that we're seeing here. And I think this growth does demonstrate that in AI-driven world, the need for collaboration at the enterprise scale, tracking, planning, managing work, it's actually increasing. And Atlassian is a mission-critical platform here, helping our customers orchestrate their teams, their agents, their workflows to really unlock that value in AI and driving real ROI and real outcomes here. And that really was the other primary driver of the outperformance that we saw in cloud, including the cross-sell that I talked about a little bit earlier.
- 支持 CEO：财务、HR、市场人员在用这些工具做技术；MCP/CLI 使用遍及各职能
Look, I can talk about it from a developer or engineering front in a number of different angles, Gregg. I know we've talked about this before. We've been long on record. You can look at the receipts I think there's getting more developers in the world in 5 years' time than there are today. The cost of building technology is going down. The amount of technology we're going to build is going to go up. I believe that will continue to be a good trend. We've said this for a number of years. Those are the signals we see in our customers, the ability to grow that is huge. I think secondly, the number of people building technology that we don't necessarily call developers is also growing. So we're seeing people in finance and in HR and in marketing, using a lot of these tools to build technology of various different forms. Again, we see that in our MCP server usage and our CLI usage coming across the board in lots of different functional areas, not just inside the engineering fraternity. And I think that's really important. From our perspective, again, the system of work is about helping companies to connect their technology teams with their business teams to have a single platform that allows both sides of the organization in a technology-driven organization, an organization where software is their key competitive advantage. It doesn't make them a technology company. It can be a bank, an insurance company, a health care company, a university. There are lots of companies for whom. Many organizations in the world, technology is a key competitive advantage. That is -- I believe that is a big growth cycle that we are continuing to participate in. Our strength is in connecting the technology team through the business team and I do believe that the developer hiring will continue to grow as we see in our customer signals.
- 支持 客户升级 TWC 或在扩展到更多知识工作者时增加 seats
I'm sure I can take that one. Look, I can tell you I've met more than 60 customers over the last 6 weeks and 5 different countries around the world and the Teamwork Graph and AI come up in every single conversation I've had. You can see from our results, those customers are continuing to expand their use of the Teamwork Collection, either upgrading to it if they're buying single applications or adding to the seats as they move into more of their knowledge workers and across the organization. And AI and the Teamwork Graph is 1 of the biggest reasons -- it's 1 of the top 2 reasons quoted by our customers for their reason to upgrade to the cloud and to upgrade the Teamwork Collection. Now the question is why and where those results come from. Again, our applications have a lot of connections to other applications around the world, right, whether that's their CRM system, whether that's their documentation knowledge systems, we have, for about 8 years, been building the Teamwork Graph as the best way to get answers across all of those applications for their users or for their agents, which increasingly ask a lot of questions in a very short period of time.
- 支持 约三分之二 Jira/Confluence 用户为知识工作者，技术与知识工作者 seat 均扩张
* **Context** TEAM’s core strategic pitch is that foundation models will commoditize, but enterprise context will not. Teamwork Graph now spans more than 200B objects and connections, with TEAM claiming up to 44% more accurate answers using 48% fewer tokens. Mgmt expects enterprises to maintain three to five major knowledge graphs and intends for TEAM to be one of them, while opening the Graph to third-party agents through MCP and CLI rather than trying to own the entire AI stack. * **Demand** Mgmt described customer demand signals as healthy and broad-based, with continued expansion in both technical and knowledge-worker seats. Roughly two-thirds of Jira and Confluence users are knowledge workers, and mgmt argued that deploying more AI agents increases the need for collaboration, planning, workflow management and coordination rather than eliminating TEAM seats. Mgmt also continues to expect the number of developers and non-developers building software to increase. * **Margins** F4Q non-GAAP OM of 36% beat Street by roughly 540 bps, and TEAM reported its first GAAP-profitable quarter with a 12% GAAP OM. FY27 non-GAAP OM of 25% looks weak against Street at 28.8%, but includes roughly four points from lapping FY26 Data Center revenue-recognition benefits and another three-point headwind from moving compensation from SBC toward cash.
来源背景：TMTB 作者转述业绩会与 callback - 支持 净新增 seats 覆盖开发者和知识工作者两类；Jira 约 2/3、Confluence 约 70% 是知识工作者
Yes. Thanks for having me, Jason. Yes, we're really pleased with the strong Q4 results to end our fiscal '26. I think what you're seeing is customers really value and understand the value of the overall Atlassian platform with the Teamwork Graph, which is a living contextual layer underlying the platform as well as our overall system of work, basically living system of record and system of work to actually drive those workflows for organizations. To Jason's point, we had a really strong quarter on the cloud side of things. And that outperformance was really driven by 2 things. It's actually consistent with what we saw in Q3. So it's good to see that consistency carry through. But it's driven by strong upgrades and cross-sell to our Teamwork Collection, which is basically customers being able to purchase the entire Atlassian platform and primarily for additional Rovo credits. We get 10x the amount of Rovo credits with the Teamwork Collection. So customers are upgrading for that additional AI capability as well as cross-sell motion into our Service Collection. So that was really great to see. All the while, we're starting to see -- continue to see strong seat expansion across our core products of Jira and Confluence, which I think highlights the importance of collaboration and teamwork and coordination in this AI era, right? Like, we've talked about for quite some time. In the AI era, the need to track, manage, plan all your work across the organization that doesn't change. And I think you're starting to see that continue to play out with strong cross-sell momentum, AI purchasing on Teamwork Collection with that strong seat expansion on our core products with Jira and Confluence. And I think it's notable that the strong seat expansion was across both software development and knowledge workers. So at our investor forum, we recently shared about 2/3 of users on Jira and over 2/3, more like 70% on Confluence are knowledge workers and non-software developers. That trend continued across this quarter where that net new seat expansion happened across both those sectors. So that continues to be really healthy across both segments.
- 支持 Service Collection 超 60% 用例在 IT 以外
Yes. So you're right. Like in Q3, we disclosed or shared that it had crossed the -- Service Collection had crossed $1 billion of ARR, growing north of 30% year-over-year. In Q4, we actually saw that accelerate as people continue to consolidate, continue to adopt Service Collection across not only their IT workflows, but also their non-IT workflows. So what's interesting, I think, it's -- is over 60% of the Service Collection use cases are outside of IT. So that's marketing teams or HR teams taking in service requests from the employees and helping drive outcomes for their employees. At the same time, we're seeing 3x increase in Rovo agentic automation in service collections specifically over a 6-month period. So all the while you're seeing customers increasingly adopt Service Collection across the organization, they're deploying Rovo agents to help triage those service requests. So when you think about an HR team, getting a ton of inbound requests from their employees, asking questions about X, Y or Z, you're able to deploy Rovo agent to be that first line of defense and triage and kind of cut out that noise, respond to employees, give them faster customer service, if you will. And then also for the more critical items that require human judgment, route those to the right expert to be able to address those questions. And so whether that's in customer service or employee service, I think we're seeing really strong Rovo adoption across those workflows.
- 口径不齐 分析师提到公司称 seat 已卖入财务/HR/营销；专家未直接确认，只说 IT 应用团队是扩张推动者、能否兑现待验证
不过，我不确定 Atlassian 能在多大程度上兑现全部承诺。从某种意义上说，我会将其与 Palantir 进行比较。Palantir 目前高度专注于企业市场，原因在于它此前曾在企业市场遭遇失败。Palantir 在政府市场表现非常出色，但这也让许多人产生疑问，不确定其产品是否真正具备所宣称的能力。我认为，Atlassian 目前正处于类似的关键阶段。
来源背景：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团；自称并非 Atlassian 专家 - 支持 Jira 65% 知识工作者，Confluence 近 70%，JSM 超 70%
We're seeing more software being built, we're seeing more seats being added through across developers as well as Non-purpose. I think one of the things that is most underappreciated about Atlassian is how diversified the set of businesses. You know, I came into Atlassian sort of thinking that it was a toolbase company simply for software engineers called Jira. And the reality is 65% of Juren users are knowledge work on software. Engineers like Confluence is even higher, closer to 70% and JSM service collection north of 70%. So incredibly diversified base of business. And I think as customers really think about how they're going to get value out of AI, it's one thing to have a a point solution product that is going to drive individual productivity. How do you actually get value and productivity across ? TI Well, that's exactly where Atlassian's Right. We've spent 20 plus years helping team collaborate, orchestrate, and that's becoming more important now, not just with people, but through people and agents together. And so as customers realize that they're adopting more of our teamwork collection, they want to be able to go wall to wall with Atlassian. We able to unlock the value of the Teamwork graph because the more teams and more individuals that they can get onto the Atlassian platform or context that they're at, the workflows that are being built across JIRA, service collection and all the context that's being captured in Confluence page that are attached to those workflows that's really creating the intelligence and the IP within APPRIZED that they can begin to unlock.
Coding agent 是加 seat 还是减 seat
判断：证据只支持"agent 还没让 seat 变少"，不支持"agent 让 seat 加速"。AI 带来的是工作量，而工作量涨得比 seat 快得多。
管理层的逻辑链是：写代码的成本下降 → 软件变多、写软件的人变多 (包括财务、HR、市场) → 规划、追踪、协作需求变多 → 更多 Jira seats。支撑这条链的数字只有一个：用 AI 编码工具的数千家客户，Jira 任务多 5%、MAU 多 5%、seat 扩张快 5%。CEO 在 2026 年 2 月主动澄清这不是 5% 的扩张率，而是比别人高 5%，并承认这批都是"增长型的前沿公司"，选择偏差很明显。"Rovo 客户 ARR 增速 2 倍"同理，是相关性。
反过来看，工作量的数字比 seat 大一个量级：经 MCP 建的 Jira 工作项是 4 倍，MCP/CLI 月活超过 100 万，MCP 调用一个季度涨 400%，而人通过网页和手机用产品"几乎没变"。这说明 AI 带来的价值主要以"不用登录的工作量"出现。这部分要变现，靠的是 credits 和用量，不是 seat。管理层在这点上也没有定论：CRO 说 agent 怎么收费"是我们和同行都在讨论的问题"；CFO 在 Citi 被问 agent 数量远超人类 seat 时，只回答了 credits。
专家这边：Atlassian 前 VP 说 AI 让客户做更多原本做不到的工作，所以 seat 不降反升，但这个观察来自他在 ServiceNow 的经历，数字是他自己说的"随便举例"。反方更多：渠道商说一个 service account 就够；Funda 的电信高管说开发者人数未来 1-2 年只增 0-5%；Co-operators 首席 AI 官说长期会减工程师；monday.com 前员工说纯 per-seat 厂商"要么转用量，要么死"。
Thank you all for joining us today. As you've already read in our shareholder letter, we closed out Q2 with very strong enterprise sales execution and incredible momentum across our business. We surpassed $6 billion in annual run rate revenue, delivered our first ever $1 billion cloud revenue quarter, up 26% year- over-year and grew our RPO 44% year-over-year to $3.8 billion. We have strong momentum across our enterprise, AI and system of work transformations, and you can see this in our numbers. Customers are choosing us for their future in bigger ways and bigger numbers than ever before. Enterprises like Cisco, Expedia, Reddit and Synchrony Financial rely on Atlassian to power their most critical business processes and workflows. Rovo surpassed 5 million monthly active users of our AI capabilities. We're seeing firsthand every day how AI is transforming the way that work gets done, and we are directly benefiting as a business. When we look at the thousands of customers in our software teams using AI code generation tools, we found that they create 5% more tasks with Jira, have 5% higher monthly active users and expand their Jira seats 5% faster than those who don't use these AI coding tools. As I've said before, AI is the best thing to happen to Atlassian and the results we're seeing today are no accident. As a long-term focused company, we're now benefiting from years of thoughtful investment across product, R&D and GTM, which have positioned us to capture this moment. These investments are creating what we believe is a truly differentiated customer experience. First, the data and domain expertise living inside our Teamwork Graph, which is now well more than 100 billion objects and connections across first and third-party tools, enables Rovo to deliver real business value that's context aware and actionable for customers across their search, chat and agentic experiences. Second, our decade-long investments in enterprise-grade security, data governance, permissioning capabilities and compliance enable every organization to securely move work forward at scale while deploying these fantastic new AI capabilities with the trust that they need. We provide a system of work built on deep integration into customer workflows with that compliance, security and support that enterprises trust built in. Lastly, our unique distribution engine enables us to seamlessly deliver those incredible experiences to over 350,000 customers, including more than 80% of the Fortune 500 and 60% of the Forbes AI 50. Customers are realizing the value of our unified system of work and want to continue to partner more closely with us than ever before. And all of this is driving our results. We closed a record number of deals greater than $1 million ACV in Q2, nearly doubling year-over-year again as enterprises are choosing to standardize on the Atlassian system of work. In less than 3 quarters, more than 1,000 customers have upgraded to our main AI monetization driver, the Teamwork Collection, purchasing more than 1 million seats to get the best AI platform and many more AI credits for their agents. As I look across our business, two things are clear. We've never been more of a strategic partner to the biggest businesses in the world, empowering their AI and future of work transformations and our momentum is continuing to grow. This gives us confidence in our road ahead and our long- term opportunity. We are truly transforming how work gets done and solving the toughest human AI collaboration challenges for our customers, and we're doing it every day. We're pushing ahead with strong conviction, and I could not be more bullish about the massive opportunities in front of us as we advance our mission to unleash the potential of every team. With that, I'll pass over to the operator for Q&A.
Koji, great question. I'll get into some specifics here. Firstly, that's obviously within sort of the Software Team and Software Collection. So the first thing I would say is to Ittai's question previously, that is a subset of our user base, right, in terms of software teams. And those are generally broad technology teams as well, right? So not just developers, but security folk and network analysts and operations teams and product managers and designers. There's a lot of different roles involved in Software team well beyond just the coding itself. But it is a subset of our total audience. And again, Service Collection had an amazing quarter in a totally non-software sense. I think what we're saying there is it's 5% higher or at least 5% higher than non-AI code generation-based companies. It doesn't mean it's a 5% expansion rate. It means they're expanding 5% higher than the rate of expansion of the other groups. So that's where you see that it's not necessarily a 5% total expand rate. You can see in our NRR and other stats, it's higher than that, right. Secondly, there's a lot of reasons for that, I believe. Firstly, these are the cutting-edge companies. These are the companies that are pushing the boundaries the most. They tend to be growth-oriented companies. So they tend to be companies that are growing, which is great. But guess who those largest companies are going to be in the future is those ones that are pushing the boundaries and driving forward in a generalized economic sense. So that's really good. Those are the leading companies for us. Secondly, yes, they're getting more efficiency. If you look at the actual delivery efficiency, coding speed is, again, 20% to 30% of the developer's job. And so you may be getting 10%, 15%, 20% improvement in the overall productivity of your organization if you have 1,000 people in R&D, something like that. But that innovation moving quicker, it doesn't mean you're finishing your road map. You're coming up with more things to do. So you're adding more tasks. You're also creating a lot more technology, a lot more software and services, which makes your architecture more complicated. It gives you more things to manage with something like Compass in terms of the different software code bases and models and pipelines and all the different data structures that you have to deliver your technology products and services as an organization as a customer of ours. And lastly, you create more complexity, right, the security and compliance of a bank, the governance functions that have to happen, the structuring and the downtime, the operating of that software. All of these things create Jira issues at large volumes, right? So if you create more software, you're going to have more management, more overhead, more collaboration. Some version of that is what we believe is happening underneath, right? You have more collaboration to do because you end up with more technology, and that's a good thing. More software in the world is a good thing for Atlassian. I think we've said that for a couple of years now. That is our belief that AI is unlocking sort of human creativity at the highest level, right? It's allowing them to create more. That means those humans have to interact and collaborate more and those created objects need to be managed, operated, maintained and that's generalized a good thing for Atlassian across software and non-software teams.
Koji, great question. I'll get into some specifics here. Firstly, that's obviously within sort of the Software Team and Software Collection. So the first thing I would say is to Ittai's question previously, that is a subset of our user base, right, in terms of software teams. And those are generally broad technology teams as well, right? So not just developers, but security folk and network analysts and operations teams and product managers and designers. There's a lot of different roles involved in Software team well beyond just the coding itself. But it is a subset of our total audience. And again, Service Collection had an amazing quarter in a totally non-software sense. I think what we're saying there is it's 5% higher or at least 5% higher than non-AI code generation-based companies. It doesn't mean it's a 5% expansion rate. It means they're expanding 5% higher than the rate of expansion of the other groups. So that's where you see that it's not necessarily a 5% total expand rate. You can see in our NRR and other stats, it's higher than that, right. Secondly, there's a lot of reasons for that, I believe. Firstly, these are the cutting-edge companies. These are the companies that are pushing the boundaries the most. They tend to be growth-oriented companies. So they tend to be companies that are growing, which is great. But guess who those largest companies are going to be in the future is those ones that are pushing the boundaries and driving forward in a generalized economic sense. So that's really good. Those are the leading companies for us. Secondly, yes, they're getting more efficiency. If you look at the actual delivery efficiency, coding speed is, again, 20% to 30% of the developer's job. And so you may be getting 10%, 15%, 20% improvement in the overall productivity of your organization if you have 1,000 people in R&D, something like that. But that innovation moving quicker, it doesn't mean you're finishing your road map. You're coming up with more things to do. So you're adding more tasks. You're also creating a lot more technology, a lot more software and services, which makes your architecture more complicated. It gives you more things to manage with something like Compass in terms of the different software code bases and models and pipelines and all the different data structures that you have to deliver your technology products and services as an organization as a customer of ours. And lastly, you create more complexity, right, the security and compliance of a bank, the governance functions that have to happen, the structuring and the downtime, the operating of that software. All of these things create Jira issues at large volumes, right? So if you create more software, you're going to have more management, more overhead, more collaboration. Some version of that is what we believe is happening underneath, right? You have more collaboration to do because you end up with more technology, and that's a good thing. More software in the world is a good thing for Atlassian. I think we've said that for a couple of years now. That is our belief that AI is unlocking sort of human creativity at the highest level, right? It's allowing them to create more. That means those humans have to interact and collaborate more and those created objects need to be managed, operated, maintained and that's generalized a good thing for Atlassian across software and non-software teams.
Yes. So the primary AI monetization motion or the primary purchasing motion for customers today is upgrading to Teamwork Collection, where you get 10x the amount of Rovo credits. We made the strategic decision about 2 years ago to thread Rovo our AI capabilities throughout the entire platform. And so with each base subscription, you get a certain allotted number of Rovo credits. And then as you approach those limits, then you upgrade to the Teamwork Collection where you get 10x the amount. And that's much more customer-friendly in this moment in time where I think customers want that predictability or value that predictability of. I get 10x the amount of credits across my entire user base at a higher price point, so we realize higher ARPU as a result instead of worrying about Jason is going to chew through x amount of credits and I have to worry about this power user. It's a fungible pool of Rovo credits spread across my entire organization at 10x the amount. So it's much more predictable and customer-friendly, and we're just meeting customers where they're at today. I think over time, consumption-based pricing or usage-based pricing becomes a bigger part of the story. We'll actually begin to enforce Rovo credit limits this year. But again, I think the primary motion you're going to see is customers choosing to adopt Teamwork Collections. So I would continue to pay attention to that because that's probably the primary monetization motion today. Customers are also -- when they adopt Rovo, they, of course, realize significantly more value. And we're seeing Rovo customers grow their ARR at 2x the rate of those that don't adopt Rovo. So actually the concept of delivering value first and foremost to customers, then we'll recoup that value back over time. That's via increased stickiness, faster seat expansion and higher up-tiering to higher value editions. Actually, what's also really interesting is we shared a bunch of MCP stats. So even if customers are using third-party agents, and tapping into the Atlassian platform via MCP, we're seeing significant growth there. So we have over 1 million monthly active users of our MCP server and Teamwork Graph CLI. So again, wanting to tap into the value of the Teamwork Graph, even if you're using a third-party agent. That's delivering significantly more value to the customer. You're seeing that in their output. So they're driving 4x the amount of Jira issued or Jira Work items created via MCP server and Confluence pages created 4x the amount via MCP server. So they're driving significantly more workflows through the Atlassian platform, which is what we want to see. And ultimately, that increased value to customers comes back to us. So exact same number. Those customers utilizing the MCP server are growing their ARR at 2x the rate of those that don't utilize MCP. So I think it highlights the mission criticality of the Atlassian platform, and it shows that it's complementary to even if you're using a third-party agent like Claude agent.
Sure. So as we a couple of things. Firstly, like Jira within customers has consistently been an orchestration layer within customers. And as we move forward in the world of AI, it's going to continue to be an orchestration layer. If that's with humans and if that's with customers using agents, firmly, I can continue to be that orchestration layer. And now as we move forward in terms of the monetization, we obviously have the opportunity to monetize in terms of agents as well. That's something, obviously, that we're looking into how we're going to monetize in terms of agents. That's a debate that we and others are having. And we will see how we're going to move forward in terms of how we're monetizing that in the future and that I know a debate that others are having as well.
And we already have some of them in our CSM product. But I think the the I think advantage that Atlassian has as a horizontal platform is the ability to stand up different types of be able to tackle different types of workloads and work that happens across organizations. And right now we see that through to more collection in terms of the ability to monetize and folks you know choosing Atlassian to be able to deploy their AI the other way as I mentioned is through some of the CLP, Sealite and MCP use cases that we're seeing across the system as well. Now for simple queries into the teamwork graph, we don't charge for that. Right ? So if you're writing a ticket, if you're trying to resolve a quikticket, but if you're tapping into the teamwork graph with much more long horizon or complex use cases that draw on that teamwork graph, we do burndown those credits for customers. Again, it's building habits, like we said, across our customer base. first driving that adoption that that the vision that I'm being breaking out which as a reminder you remetering as of December. Yeah, that's right.
**BR** [00:02:06]: I've been selling AI as part of workflow products for the better part of 10 years, the first iterations of AI, of course, being machine learning-based, and then that's changed over the last years to agentic and generative AI capabilities. During that time, what I found is it's hugely valuable to customers and to those workflow solutions because, really, what customers are looking to do is have their problem solved, and AI can come in and actually make workflow solutions more efficient and effective. I'd say one other thing, and this will come up probably in some of the other discussion points today, what surprised me as I've been selling the AI software to customers, in the entire time, I thought the seat counts were going to go down. While I was at ServiceNow, in particular, when we were talking, we expected that seat counts would go down as people bought more AI. The reality was the opposite, and the reason that it was the opposite is because, almost all of the customers that we were selling to in Atlassian is similar in this regard, all of them are barely touching the work that they should be doing anyway. At most, prior to AI, they were getting to about 10% of the potential work that they had on their plates. AI is helping them get to, what, 30%, 40% of the work now. There's still a lot of untouched work, so it doesn't mean that seat counts go down. It certainly means that workflow software can be further made more efficient than it is and, therefore, really a benefit, to your point, to Atlassian and to other SaaS companies as well.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20
**BR** [00:05:03]: Certainly, it's my opinion, but I don't see revenues and seat counts necessarily going down, and part of the reason is the… I don't know if it's a secret or not, but AI creates work. The more efficient and effective you get… Let's just take an example which fits right in Atlassian's area. Developers, let's say they used to be able to develop two applications a quarter. This team used to do two applications a quarter. With AI, they can double their output, again, just making these numbers up. They can double their output. Who's going to support those four applications now? It used to be the same team supporting two applications. Now, they're supporting four applications, all of which… They're now being sold to customers. It means you have to have more salespeople actually selling to more customers. You have to have more support people actually supporting more customers. The same development team has to be more effective and efficient to be able to support now four applications as opposed to two.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20
从我个人了解到的情况来看，实际情况介于两者之间。AI 提效并没有导致研发人数（developer headcount）同比例缩减；更贴切的描述是工程团队人数保持相对平稳（relatively flat），处于低个位数增长（low single-digit growth），而产出增长得快得多。需要明确的是，生产力提升（productivity gain）并不会 1:1 转化为端到端交付提升（end-to-end delivery gain）。但对公司而言，AI 编程工具确实让单个开发者产出了更多代码、PR 和变更。很多代码现在都是 AI 写的，占比可达 60%、70%、80% 甚至 90%，但这反而极大地增加了下游工作量。公司现在面临更多的 code review、CI/CD 运行、安全扫描（security scanning）、测试、文档、工单（tickets）、部署以及可观测性需求。到了2026年，我认为大约 80% 到 90% 的代码都由 AI 生成，但同时也带来了许多工作流问题。 对于像 Atlassian 这样的厂商（Jira 和 Confluence），不能简单地将其收入按开发者席位（developer seat）来建模。实际上，Jira 和 Confluence 本身正在从单纯的人类开发者席位工具，转变为融合人类与 AI Agent 工作流的记录系统（system of record）。Jira 正在明显向管理人类与 AI Agent 的规划、治理和度量层（planning, governance, and measurement layer）推进。 而像 Ansible 这类工具则有所不同。AI 产生了更多软件工程基础设施变更后，公司反而需要更多的自动化（automation）、配置管理（configuration management）、策略执行（policy enforcement）和可重复部署（repeatable deployment）。
来源背景：Strategy Executive，美国头部电信运营商（前任职）
JL: But I mean, you know, your question is, we will never get to 0 developers. I think the value or the vision is that we can either, well, certainly over the next 24 months, it would be to create capacity amongst our developers to tackle more complex, more sophisticated application development, coding, testing capabilities, machine learning development. And utilize either Cognition or Factory to support the lower risk, lower complexity types of activities. Over a longer term horizon, we would expect that both factories' ability to support more complex development improves. And then that creates an opportunity for us to decrease engineer, architect, scientist headcount.
来源背景：Julian Lustig，Co-operators VP Digital & Data Platforms / Chief AI Officer
Yes. So Jira is extremely valuable in the context of any company that's deployed it, not because it's a project management tool, but because it's a workflow tool. So Jira has access to so many different workflows along how work happens. For example, what's the customer reporting in terms of issues and what -- how is that issue getting resolved? What's the full request that went into production? What was the incident in production and how did we manage that incident? All of those workflows get captured in Jira. So in the context of AI-induced software development, what changes is a lot of those cycles of software development now are getting compressed. So earlier where you used to have one developer, one designer, one program manager, one DevOps engineer and do all these handoffs, a lot of that power is now getting diffused and concentrated with generalists, so to speak. So Jira, therefore, becomes less of a project management tracking tool in terms of, let me write down what's on the road map, but more an orchestration engine. So irrespective of what kind of software you're building, as long as it's sophisticated enough and it's used in the real world, it will involve more than one person doing the job. And how do you coordinate that work when there's more than one person? The unit of coordination may not look like a bug or a story point. The unit of coordination might look like AI agents kicking off workflows. So now how do you orchestrate these multiple agents working together? Where do they start and stop? How do you push notifications and control all of those? And Jira is the perfect point for that. It's the perfect surface area for it because it has visibility across all of your teams. So we see Jira very much going in that direction, and that's what we keep building inside of.
展开全部 80 条原文 (按时间)
- 支持 CEO 把 agent 定位为'给团队加人'，而不是替代人（定性，未谈 seat 计费）
So what's very different about Rovo is the way that an agent interacts inside the software, as another colleague in your company would. That makes it much more familiar to the usage pattern than some of the other approaches to AI, which has gone pretty great reactions from customers. And the ability to iterate, as I said, between humans and agents and back and forth and obviously multiple times, 2, 3, 4 loops to get a task done, that also resonates really strongly with our approach. It feels more like adding extra people to your team, as we would phrase it, than it does someone sort of replacing what you're doing, which creates huge positivity in our customer base. Generally super early in our journey here, but a huge level of customer excitement. And our goal is to get that into as many hands as possible. I think we're making pretty good traction.
- 支持 有客户公开谈 AI 影响员工规模，但他们的 license 仍在扩张
And that is a great thing for Atlassian because what that means is as more software gets created, more people get involved in this entire creation process. And fundamentally, we are in the business of teamwork. So whether it's humans collaborating with other humans, humans collaborating with AI, all of those workflows need management, and that's where we come in. In fact, there's a lot of customers who have openly talked about how AI is going to impact their teams and their workforce, and we continue to see their licenses expand. Why? Because our tools serve every single profile that's involved in creation and not just the people writing code. In fact, the code generation tools, which I think perhaps command more share of the news than is, one, is 20% of developers' time on average is consumed by coding, the remaining 80% is how do I work with the designer, how do I work with the customer support person to resolve bugs? So what about the remaining 80%? That's what we're focused on. How do we make that efficient? And I think there is tremendous opportunity and upside there for Atlassian's business overall to capture those workflows.
- 支持 AI 让不会写代码的人也能做软件，参与创造的人变多，对 Atlassian 有利
So we've managed to increase the usage and adoption by 50% just over the last quarter. Business-wise, why this is exciting for Atlassian is that AI as a technology wind, as a technology trend basically means that more software gets created. So earlier, you needed to be a highly technical person writing code in order to create software, whereas now that same power of creation is available to everybody. So you don't have to have written a line of code in the past, but you're now able to build applications, build software despite what your background is. And that is a great thing for Atlassian because what that means is as more software gets created, more people get involved in this entire creation process. And fundamentally, we are in the business of teamwork. So whether it's humans collaborating with other humans, humans collaborating with AI, all of those workflows need management, and that's where we come in. In fact, there's a lot of customers who have openly talked about how AI is going to impact their teams and their workforce, and we continue to see their licenses expand.
- 支持 高技术 SMB 和非技术 SMB 的扩张率一致（没给数字）
So I think on both sides, we are seeing this. And one concrete piece of evidence that we see is you hear a lot about smaller companies, AI-native companies, starting out with really small numbers of people and being able to do a lot more with fewer people. Because we serve 300,000 customers, we have a lot of SMBs within our customer base as well. And when I compare some of the small, medium businesses that are highly technical, compared to the small, medium businesses that are not necessarily in technical areas, I see the expansion rate continue to be consistent across both.
- 反对 JSM 用 AI 自动处理工单，可以不用人工处理，可能压缩服务团队的人力
And this is what we see with a lot of our customers as well. So I think as we make each of these creation teams a lot more powerful, more software gets created, and therefore, we will see more and more usage of our products. But going back to your point of there are some things that are profit centers and some things that are cost centers; at Atlassian, we make it a point to serve all those kinds of things. So the kind of behavior that we see is that as we provide things like customer service management, like Jira Service Management for automatic ticket handling such that you don't have to have a human handling, same kind of questions that come up, IT help desk, new hire onboarding questions, we're able to go deflect that and use AI to automate that. We see a lot of those teams getting a lot more productive and they're willing to really pay for that productivity step up, right? And that's a good thing. And on the revenue generation side, the creative side, we're able to supercharge the teams that can create these new things, can deploy their creativity to create revenue-generating objects.
- 支持 科技创业公司（最拥抱 AI）的 seat 扩张模式和其他行业 SMB 一致
I think what's also interesting, when we look at our seat growth, our seat growth has been very consistent over the past 4 quarters. So we continue to see really healthy seat growth. What's been also really interesting is when you cut it by tech startups, right? You would imagine that tech start-ups tend to be the most tech-forward companies. Their seat expansion patterns have been consistent with the seat expansion patterns that you see with SMBs across other industries. So that's a really encouraging sign for me, which again, I think, shows you our ability to serve not just probably developers, but also other types of users. And then earlier, you mentioned our focus on enterprise, right? And so while we certainly drive a lot of our growth via seat expansion, we also have developed different growth vectors in terms of upgrading people to premium and enterprise additions upselling and focusing on these enterprises, cross-selling across the broader portfolio of products and then pricing, right, capturing it via consumption pricing or perhaps per seat pricing over time. So I think we feel really good about kind of the culmination of these different vectors and our ability to drive durable growth no matter how the world continues to evolve with Gen AI.
- 支持 AI 会从用户数上打开 TAM，不懂技术的人也能做软件元素
How do you understand what customers want? How do you manage all this and iterate on what to build next? These are the problems that humans will have to solve. I think these technologies also open up the TAM in terms of the number of users, right? Like I actually built an agent last week, I don't have any technical skills. I think you're going to see a lot more people with the ability to create software elements. And again, you'll then have more and more business people that will have to interact with all the software that's being created, right? Again, how do you hire for it? How do you manage the cost of this? There'll be all these new problems where people have to come together to solve these new problems. And I think that's a good problem for Atlassian because that's what our products serve. I also mentioned earlier that we've introduced consumption-based pricing elements within our products, and that's a way to kind of capture value, not just through a per seat basis.
- 支持 SMB 科技创业公司过去一年的 seat 扩张模式稳定，和其他行业同等规模的 SMB 一致
Yes. And again, I think it's important to remember that fundamentally Atlassian is a collaboration company, right? Like all of our products facilitate collaboration. And so as the output increases of software because no doubt it will, there become all these new problems that humans have to collaborate together on and solve and Brian echoed all of that, right? Like how do you manage all this new software? How do you sell it? How do you get feedback from what customers want so that you can iterate on that software? Again, there's an abundance of new problems that get created. I think it's also important to remember that already today, 50% of our seats are technical in nature, so that's developers, engineering, IT. And already, we serve 50% that are nontechnical. So those business users, and we'll continue to push more in that direction. When you actually look at some of these SMB tech start-ups that you would expect to be much more kind of embracing or forward tech embracing, when you look at their patterns of seat expansion, it's been consistent for the past year. So that's good. And then also when you contrast that against SMBs of similar size in different industries, it's very consistent. So you don't see any deviation in kind of seat expansion patterns. Continue to see very healthy seat growth, continue to see absolute seat growth. I think the other interesting anecdote is when we look at some of these CEOs or people that make kind of bombastic proclamations, when you look at the underlying customer data, we actually see seat expansion within a lot of these companies. And so it's -- I think we feel really good as we continue to see seat growth because of all these initiatives and projects that continue to develop, as Brian said, but also as we kind of continue to push more to serving these nontechnical users.
- 支持 发出“大胆言论”的 CEO 所在公司，底层数据显示 seat 实际在扩张
Yes. And again, I think it's important to remember that fundamentally Atlassian is a collaboration company, right? Like all of our products facilitate collaboration. And so as the output increases of software because no doubt it will, there become all these new problems that humans have to collaborate together on and solve and Brian echoed all of that, right? Like how do you manage all this new software? How do you sell it? How do you get feedback from what customers want so that you can iterate on that software? Again, there's an abundance of new problems that get created. I think it's also important to remember that already today, 50% of our seats are technical in nature, so that's developers, engineering, IT. And already, we serve 50% that are nontechnical. So those business users, and we'll continue to push more in that direction. When you actually look at some of these SMB tech start-ups that you would expect to be much more kind of embracing or forward tech embracing, when you look at their patterns of seat expansion, it's been consistent for the past year. So that's good. And then also when you contrast that against SMBs of similar size in different industries, it's very consistent. So you don't see any deviation in kind of seat expansion patterns. Continue to see very healthy seat growth, continue to see absolute seat growth. I think the other interesting anecdote is when we look at some of these CEOs or people that make kind of bombastic proclamations, when you look at the underlying customer data, we actually see seat expansion within a lot of these companies. And so it's -- I think we feel really good as we continue to see seat growth because of all these initiatives and projects that continue to develop, as Brian said, but also as we kind of continue to push more to serving these nontechnical users.
- 支持 开发者效率提高会带来更多项目、更多使用，seat 不会下降
Yes. Good question. So for us, we see this as actually really being a tailwind and something that's going to help our business. So obviously, developers are incredibly important to us and are a large portion of the seats, overall seats for us. Now when you think of developers, obviously, a portion of their time is spent literally hands on keyboards in terms of writing code. But actually, only 1/3 of their time is really spent hands on keyboard. The rest of their time is spent collaborating with team members and managing the projects. And they will be doing that predominantly via our tools, Jira, Confluence, et cetera, which is great for us. Now as new tools like you highlighted, are introduced, those tools are going to be helping them actually code more efficiently and effectively bending the curve of productivity for these developers. And that's good news for them and good news for the projects that they're working on because that allows them to do projects more efficiently and, in theory, do more projects. But as they're doing those projects, it doesn't change our world because they're actually still managing all of those projects through the same pane of glass, which is Jira and Confluence. And if anything, they're actually doing more projects. So instead of seats potentially going down, for us, what we see is developer productivity increasing for projects that we're working on and more utilization of our license -- of our solutions and not an impact that may be what you were alluding to in terms of a decline of seats. So for us, we actually see a tailwind here. And maybe I'll let Martin comment on -- maybe you want to talk about some of the startups?
- 支持 更多项目带来更多使用，更多使用带来更多 seats（依据是 Duffy 自己团队的经验）
Yes. And like the more -- even for myself, but our developers who are delivering projects for me who are utilizing some of these tools as well. The number of projects in the last few months have been steadily increasing. And then my utilization and other people within my team, utilization of Jira and other tools has now been steadily increasing, quite frankly, because the number of projects that we have been increasing because the deliverables that they have been increasing as well. So I think as more and more tools and products, software come to market that increases their productivity is actually going to help us because it's going to be driving more efficiency, which will drive more projects, more projects drives more utilization, more utilization will drive more seats, which is good business for us.
- 支持 CEO：客户接入 code gen 工具后对产品使用没有变化，用户增长依然健康（定性，无数字）
Sure, Keith. Good to hear from you. Let me just start by saying we're not seeing any impact from this on any numbers that we have. So expansion rates, growth rates, adoption rates of our technical products. And again, the technology teams and the software business is a part of our business alongside service strategy and broad collaboration. But we're not seeing any impact from it. So we're being very clear about that. When customers integrate code-generating AI tools, there's no change in their usage of our products, and we continue to see really healthy user growth. In fact, we even have some early data that shows that when people do connect cogeneration tools, the business users in those geo instances is actually growing faster than they are for the customers that don't have those connected, right, demonstrating, we think, Jira continue to be the center gravity in how work gets done. I think you kind of have to look at a high level. Let me give you some assumptions that I have or some beliefs.
- 支持 早期数据：接入 code gen 工具的客户，业务用户增长快于未接入的客户（未量化；'geo' 应为 Jira 的转录错误）
But we're not seeing any impact from it. So we're being very clear about that. When customers integrate code-generating AI tools, there's no change in their usage of our products, and we continue to see really healthy user growth. In fact, we even have some early data that shows that when people do connect cogeneration tools, the business users in those geo instances is actually growing faster than they are for the customers that don't have those connected, right, demonstrating, we think, Jira continue to be the center gravity in how work gets done. I think you kind of have to look at a high level. Let me give you some assumptions that I have or some beliefs. Do I think there will be far less developers in the world 5 years from now? No, I don't think so. I think there'll be far more and far more people creating software in other functions, right, whether they're in finance or HR or marketing, there's going to be a lot more people creating software. Do I think developers roles change and augment with these tools? Yes. Develops as use of that. This is a big change. It's super productive. We see that internally, but there's still a lot of -- awful lot of work to do, right? Do I think more people will be creating software? Yes. Do I think more software will be made in the world? Yes, I think there'll be far more software and technology made that they need to be managed, right? Far more people creating software and far more technology significantly expands our opportunity. Software creation is becoming more efficient. Cost of software building going down. This is all really good for Atlassian and our workflow tools.
- 支持 Atlassian 内部用了多款 code gen 工具，仍在大量招聘工程师
We have Rovo Dev, which sits on top of SWE-bench, full bench -- benchmark time at the moment. We use a whole series of other cogeneration tools. We have 3 or 4 in operation with over 1,000 now each work, lets us create software faster, let's just create better software and let's just create more software. And yes, we're still hiring full lot of engineers and developers with the growth of the business. So I just want to be clear that we are not seeing that in our numbers. We don't believe in that in terms of our collaboration things -- collaboration products. Again, Atlassian is here to solve people problems, not technology problems.
- 支持 谁能创造软件在快速变化，非技术人员也能创造软件，这对 Atlassian 有利
The belief around what does AI mean for our platforms because there's one school of thought which worries a lot about, oh, will AI put developers out of work and what happens to seats, et cetera. And then there's another school of thought around what is actually happening on the ground and how is work evolving and what role can companies play in that. At Atlassian, fundamentally, we think about it as AI is going to cause more software to be created. And more software being created is a good thing for us as a business because we are in the business of serving teams that build software. Now the way that teams build software is changing rapidly, who gets to create software is changing rapidly. So it's no longer the technical developer, but also the nontechnical consumer that can create software. So for Atlassian, that's a good thing in terms of how do we serve these nontechnical customers in addition to the traditional technical developer that creates software. The good thing for us is that well over half our business has been serving nontechnical users over the last couple of decades. And so for us we think about the opportunity is really in terms of how do we build these brand-new workflows that serve these nontechnical audiences, which is what we do with Rovo Studio, with Rovo Agents. And how do you really unlock knowledge from various different silos and bring it in service of these new people who are creating a lot of software. In terms of the risk that creates, there is a risk that -- given all the fraud in the market, there is a risk of fragmentation where enterprises end up trying various different tools. But we see it mostly as a cycle. So as people try out new different tools and see what's in it for them and how to use it, typically, enterprises come back to consolidating on a platform and the fact that we can serve different kinds of teams, customer support teams, developer teams, marketing teams, HR teams, we are able to provide that one consolidated platform at Atlassian. So that's the reason we keep focusing on what is that plane of orchestration we can deliver for enterprises to consolidate on one connected data platform.
- 反对 AI 压缩开发周期，原来多个角色交接的工作集中到通才身上
Yes. So Jira is extremely valuable in the context of any company that's deployed it, not because it's a project management tool, but because it's a workflow tool. So Jira has access to so many different workflows along how work happens. For example, what's the customer reporting in terms of issues and what -- how is that issue getting resolved? What's the full request that went into production? What was the incident in production and how did we manage that incident? All of those workflows get captured in Jira. So in the context of AI-induced software development, what changes is a lot of those cycles of software development now are getting compressed. So earlier where you used to have one developer, one designer, one program manager, one DevOps engineer and do all these handoffs, a lot of that power is now getting diffused and concentrated with generalists, so to speak. So Jira, therefore, becomes less of a project management tracking tool in terms of, let me write down what's on the road map, but more an orchestration engine. So irrespective of what kind of software you're building, as long as it's sophisticated enough and it's used in the real world, it will involve more than one person doing the job. And how do you coordinate that work when there's more than one person? The unit of coordination may not look like a bug or a story point. The unit of coordination might look like AI agents kicking off workflows. So now how do you orchestrate these multiple agents working together? Where do they start and stop? How do you push notifications and control all of those? And Jira is the perfect point for that. It's the perfect surface area for it because it has visibility across all of your teams. So we see Jira very much going in that direction, and that's what we keep building inside of.
- 反对 协作单元可能变成 AI agent 发起的工作流，而不是人
Yes. So Jira is extremely valuable in the context of any company that's deployed it, not because it's a project management tool, but because it's a workflow tool. So Jira has access to so many different workflows along how work happens. For example, what's the customer reporting in terms of issues and what -- how is that issue getting resolved? What's the full request that went into production? What was the incident in production and how did we manage that incident? All of those workflows get captured in Jira. So in the context of AI-induced software development, what changes is a lot of those cycles of software development now are getting compressed. So earlier where you used to have one developer, one designer, one program manager, one DevOps engineer and do all these handoffs, a lot of that power is now getting diffused and concentrated with generalists, so to speak. So Jira, therefore, becomes less of a project management tracking tool in terms of, let me write down what's on the road map, but more an orchestration engine. So irrespective of what kind of software you're building, as long as it's sophisticated enough and it's used in the real world, it will involve more than one person doing the job. And how do you coordinate that work when there's more than one person? The unit of coordination may not look like a bug or a story point. The unit of coordination might look like AI agents kicking off workflows. So now how do you orchestrate these multiple agents working together? Where do they start and stop? How do you push notifications and control all of those? And Jira is the perfect point for that. It's the perfect surface area for it because it has visibility across all of your teams. So we see Jira very much going in that direction, and that's what we keep building inside of.
- 支持 使用 code gen 工具（剔除 Rovo Dev）的客户 Jira paid seats 扩张率比未使用者快 5%，项目数多 20%+
The statistic you gave. We look at a cohort of our customers that were using co-generation tools, GitHub Copilot, [indiscernible], Cursor. We excluded Rovo Dev on purpose, so as to remove bias of our own sort of customer base. And those customers using those cogeneration tools were expanding their paid seats on Jira at a rate 5% faster than those who didn't. They were managing more than 20% more projects than those that didn't. And importantly, all of those 3 and others are working with us in partnership to bring their agents into Jira, into the business processes and workflows that they are using because that's where customers are doing work. The other anecdotes that we might have. Look, we've given a lot of statistics here, right? The Teamwork graph is phenomenal. It's up over 100 billion objects in connections and continuing to grow at a really incredible clip. Our AI interactions are up. I think it's almost 150% in the last 6 months.
- 支持 CEO 重申 5 年后开发者会更多，写软件的人更多
Thanks, Alex. Look, I'm well on record at this point as saying, I firmly believe there will be more developers in 5 years' time. There'll be far more people creating software and a far greater amount of technology in the world, which is great for all of us. And that significantly expands Atlassian's opportunities. Why do I have such conviction? Maybe important to note. We have the best data. We have amazing visibility right now, 300,000 customers, 80% of the Fortune 500, 60% of the Forbes AI 50 Atlassian customers, right? We're mission-critical and central to their business processes. We have tens of millions of developers, engineers, product managers, designers that use our applications across millions of teams.
- 支持 （引用帖）AI 降低任务摩擦会提升需求，企业会招更多人——更多 seat
Rebuttal: History shows that when you reduce the friction of a task, demand for that task increases. As AI makes employees more productive, companies often hire more people to attack higher-order problems. More seats! Furthermore, top-tier SaaS is already pivoting to consumption/outcome-based pricing (per conversation, per resolution). This allows them to capture the value of the digital labor directly, likely expanding the TAM beyond what human "seats" could ever cap out at. 6. Legacy SaaS will struggle to transition to outcomes Rebuttal: They are the only ones capable of selling outcomes. To sell an "outcome," you must be able to measure the "before" and "after." The System of Record holds the historical truth. ServiceNow knows exactly how long an IT ticket took to resolve in 2023 vs 2026. They are in the absolute best position to prove value and charge for the outcome because they own the ledger that defines what an "outcome" is. A startup has no baseline data to prove they delivered the result.
来源背景：VIC 作者 jgalt（买方）及其引用的 X 帖 - 支持 数千个使用 AI code gen 工具的软件团队客户：Jira 任务多 5%、MAU 高 5%、seats 扩张快 5%
Enterprises like Cisco, Expedia, Reddit and Synchrony Financial rely on Atlassian to power their most critical business processes and workflows. Rovo surpassed 5 million monthly active users of our AI capabilities. We're seeing firsthand every day how AI is transforming the way that work gets done, and we are directly benefiting as a business. When we look at the thousands of customers in our software teams using AI code generation tools, we found that they create 5% more tasks with Jira, have 5% higher monthly active users and expand their Jira seats 5% faster than those who don't use these AI coding tools. As I've said before, AI is the best thing to happen to Atlassian and the results we're seeing today are no accident. As a long-term focused company, we're now benefiting from years of thoughtful investment across product, R&D and GTM, which have positioned us to capture this moment. These investments are creating what we believe is a truly differentiated customer experience.
- 口径不齐 CEO 澄清：不是 5% 的扩张率，而是扩张率比其他组高 5%；且只是软件团队这个子集
But it is a subset of our total audience. And again, Service Collection had an amazing quarter in a totally non-software sense. I think what we're saying there is it's 5% higher or at least 5% higher than non-AI code generation-based companies. It doesn't mean it's a 5% expansion rate. It means they're expanding 5% higher than the rate of expansion of the other groups. So that's where you see that it's not necessarily a 5% total expand rate. You can see in our NRR and other stats, it's higher than that, right. Secondly, there's a lot of reasons for that, I believe.
- 口径不齐 CEO 承认 cohort 多是前沿、增长型公司（选择偏差）
So that's where you see that it's not necessarily a 5% total expand rate. You can see in our NRR and other stats, it's higher than that, right. Secondly, there's a lot of reasons for that, I believe. Firstly, these are the cutting-edge companies. These are the companies that are pushing the boundaries the most. They tend to be growth-oriented companies. So they tend to be companies that are growing, which is great. But guess who those largest companies are going to be in the future is those ones that are pushing the boundaries and driving forward in a generalized economic sense. So that's really good.
- 支持 CEO 解释机制：编码只占开发者工作的 20-30%，效率提升带来更多任务、更复杂的架构和合规，产生更多 Jira issue
And so you may be getting 10%, 15%, 20% improvement in the overall productivity of your organization if you have 1,000 people in R&D, something like that. But that innovation moving quicker, it doesn't mean you're finishing your road map. You're coming up with more things to do. So you're adding more tasks. You're also creating a lot more technology, a lot more software and services, which makes your architecture more complicated. It gives you more things to manage with something like Compass in terms of the different software code bases and models and pipelines and all the different data structures that you have to deliver your technology products and services as an organization as a customer of ours. And lastly, you create more complexity, right, the security and compliance of a bank, the governance functions that have to happen, the structuring and the downtime, the operating of that software. All of these things create Jira issues at large volumes, right? So if you create more software, you're going to have more management, more overhead, more collaboration. Some version of that is what we believe is happening underneath, right? You have more collaboration to do because you end up with more technology, and that's a good thing. More software in the world is a good thing for Atlassian. I think we've said that for a couple of years now. That is our belief that AI is unlocking sort of human creativity at the highest level, right? It's allowing them to create more. That means those humans have to interact and collaborate more and those created objects need to be managed, operated, maintained and that's generalized a good thing for Atlassian across software and non-software teams.
- 支持 使用 coding 工具的客户 seat 增加 5%
Sure. So I would say, firstly, we're not seeing a reduction in the number of seats, right? The conversations that we're having with customers are not around that. We're not seeing that in the conversation. We're not seeing that at the time of renewal with our customers. Instead, the conversations that we're having with our customers are very different. I'll give you an example. Just earlier today, I spoke with a large U.S.-based consumer tech company that was talking to us about a long 7-year term agreement that they wanted to enter into us, and this is a company that has over 100,000 seats with us both in the developer and the IT space. We firmly believe that having the ability for developers and nondevelopers to create software leads to more software. And same as what Aaron said, this means that there will be, one, the hiring of more developers and the creation of more software. And more software leads to more projects. What we have found, in fact, is that we see -- and for companies that are utilizing these cogeneration tools, there actually is a 5% increase for us in seats. And a 5% increase in fees obviously leads to expansion with our customers. And I know that's contrary to the headlines that you just gave there, but this is having a tailwind for us rather than a headwind. And that is what is leading to the headlines that we made around $1 billion -- our first $1 billion quarter in Q2, the 600 customers that are now paying us over $1 million and an RPO growth that is north of 40%. If we were facing all of these headwinds, we wouldn't be able to deliver numbers consistently like this quarter-over-quarter. So if anything, like I said at the beginning, AI is actually a tailwind rather than a headwind for us.
- 支持 更多软件带来更多项目，更多项目带来更多对 Atlassian 的需求
Sure. So one, we firmly believe that AI is a great thing for the world. We believe that AI is a great thing for Atlassian and a great thing for our customers. Just last week, we announced that we now have agents which are available in Jira, available in Jira Service Management, which means that our customers can now assign tasks to agents. Therefore, that means that developers, business leaders, et cetera, will have to spend less time on certain tasks and the agents will be able to do the work. That obviously is going to mean that it's going to free up an incredible amount of time for those individuals and allow them to spend more time being more productive in other areas. That is obviously going to unlock a huge amount of time for those developers as well. And ultimately, that's going to be a game changer. Developers actually only spend around 30% of their time coding, in fact, which I think always come as a huge surprise to everybody. And as we move forward, we're going to look at more and more ways that we can continue to embed more agentic features into our products. If anything, what's going to happen as we move forward, we firmly believe that all of this is going to lead to more software. More software leads to more projects. More projects leads to more demand for Atlassian. That's what we've already seen happen and pretty sure that that's going to be the -- what the future is going to look like as we move forward as well.
- 口径不齐 Jira/JSM 可以把任务分配给 agent，开发者和业务负责人花在某些任务上的时间会减少
Sure. So one, we firmly believe that AI is a great thing for the world. We believe that AI is a great thing for Atlassian and a great thing for our customers. Just last week, we announced that we now have agents which are available in Jira, available in Jira Service Management, which means that our customers can now assign tasks to agents. Therefore, that means that developers, business leaders, et cetera, will have to spend less time on certain tasks and the agents will be able to do the work. That obviously is going to mean that it's going to free up an incredible amount of time for those individuals and allow them to spend more time being more productive in other areas. That is obviously going to unlock a huge amount of time for those developers as well. And ultimately, that's going to be a game changer. Developers actually only spend around 30% of their time coding, in fact, which I think always come as a huge surprise to everybody. And as we move forward, we're going to look at more and more ways that we can continue to embed more agentic features into our products. If anything, what's going to happen as we move forward, we firmly believe that all of this is going to lead to more software. More software leads to more projects. More projects leads to more demand for Atlassian. That's what we've already seen happen and pretty sure that that's going to be the -- what the future is going to look like as we move forward as well.
- 支持 通过 MCP 用 Cursor/Claude Code 的客户：Jira 任务多 5%，MAU 多 5%，seat 扩张快 5%
Yes. Look, I think we continue to grow with those technical users and nontechnical users. So across both vectors, which is great to see. I think -- two things I would first share. It's like one, we look at those customers that are using these co-gen tools. So again, trying to address some of the fears out there. When we look at those customers that use a cursor that use a cloud code via MCP across thousands of customers, so not an insignificant number. Those customers are driving 5% more tasks through Jira. It's leading to actually 5% greater monthly active usage because to your point, more people have to collaborate to manage this increased complexity or throughput of work. And then ultimately, it's driving those customers to expand their seats 5% faster than those customers that don't use these co-gen tools. So it's really interesting. And again, it kind of backs up what we've always believed in, but we thought it was important to share with the market. And then number two, I think we get questions about, okay, what about these hot new AI start-ups. So we picked the nonsubjective list. The Ford's AI 50. I didn't want to cherry pick some stats. I just drew that over to my data science team and said how many of these -- the hottest startup -- AI start-ups in the world are our customers today? 60% are already our customers today. And they're all growing their seat counts much faster than the overall corporate rate, which is probably not too surprising, but again, it's like -- so we feel good about our kind of position in an AI world, especially even with these hottest companies that investors may have questions about. Hey, do you know those companies work in a different way? It seems like we're well positioned even with these companies.
- 支持 Forbes AI 50 中 60% 是客户，seat 数增速远高于公司整体
Yes. Look, I think we continue to grow with those technical users and nontechnical users. So across both vectors, which is great to see. I think -- two things I would first share. It's like one, we look at those customers that are using these co-gen tools. So again, trying to address some of the fears out there. When we look at those customers that use a cursor that use a cloud code via MCP across thousands of customers, so not an insignificant number. Those customers are driving 5% more tasks through Jira. It's leading to actually 5% greater monthly active usage because to your point, more people have to collaborate to manage this increased complexity or throughput of work. And then ultimately, it's driving those customers to expand their seats 5% faster than those customers that don't use these co-gen tools. So it's really interesting. And again, it kind of backs up what we've always believed in, but we thought it was important to share with the market. And then number two, I think we get questions about, okay, what about these hot new AI start-ups. So we picked the nonsubjective list. The Ford's AI 50. I didn't want to cherry pick some stats. I just drew that over to my data science team and said how many of these -- the hottest startup -- AI start-ups in the world are our customers today? 60% are already our customers today. And they're all growing their seat counts much faster than the overall corporate rate, which is probably not too surprising, but again, it's like -- so we feel good about our kind of position in an AI world, especially even with these hottest companies that investors may have questions about. Hey, do you know those companies work in a different way? It seems like we're well positioned even with these companies.
- 支持 MCP 访问里有写入和编辑，带来更多工作流，进而带来更多协作用户和升级
We continue to believe that the open ecosystem approach is an important one. If customers want to tap into Atlassian products, like that's going to be an important one because they should have that freedom to do so. We want to encourage that. What's important, when we look at the MCP data, customers are writing and editing within our products. That's an important aspect rather than just purely reading because that means that we're driving more and more workflows through the platform, and that begets more and more users to collaborate on those workflows and then that also drives upsell opportunity or upgrade potential. So we encourage that type of access into our platform. It is an area that we continue to monitor. I think we look at how the industry continues to evolve about how to maybe monetize this kind of access platform. But again, that open philosophy is a really important one for us. And philosophically, that's something that we believe in.
- 反对 公司内部用 Cursor、Claude、Rovo Dev 后开发者效率约提升 25%，同时在放缓招聘
Again, a lot of it is just thinking about how do we reinvent certain workflows to utilize and leverage AI. I think, like one of the interesting ones is we get asked a lot about our R&D team, right? Are you using some of these co-gen tools? I think what people miss is coding is only about 30%. We do surveys. You can look at third-party surveys. Coding's only 30% of the developer's time today, hands on keyboard development time coding time. So where we try to apply our AI capabilities is on that other 70% as well. Like how can you speed up some of the other 70% of what they do in collaborating, how do you figure out what we're trying to build? On the whole, we use cursor. We use cloud. We use a lot of tools, Rovo dev in the mix. Kind of the net-net of that is our developer is about 25% more efficient, which I think shows there's a lot of opportunity.
- 支持 目前客户因 AI 而增加资深开发者以审核 AI 代码，未受大裁员影响
SD: I think it will approach what my expectations were for the year. And I was expecting 30% to 35% growth for Atlassian in this year. Now, a lot of it is going to... the non-PubSec stuff is very tied to AI and AI expectations. We have not been affected by any major layoffs. They haven't hit our customers yet. I anticipate some will, but the majority of our customers are seeing the need to increase developer headcount because of AI and that there's so much productivity enhancements around things like Claude Code and whatever they pick for their corporate AI solution that they need more seasoned developers to vet what is happening and to shepherd or guide AI usage in their products. I do see eventually that we're kind of... I've used this analogy before, it's a big bell curve and we're climbing the front end of that bell curve now, and as we go up the bell curve, we need more eyes on what's happening, and that means we need seasoned software developers to understand where AI can really help and to vet what AI produces. Is it really helpful? Is it something that's maintainable? Is it something that is going to work? There's no... no vulnerabilities, no security issues in it. As we are producing 10 times the amount of code, maybe even more, due to AI, we need to make sure it's all kosher. And that means senior software developers. Once we hit the top of that where we are fully
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 反对 AI 编码工具通过 MCP 以单一服务账号操作 Jira，不需要每人一个 seat
SD: Yeah, and basically this is all because of AI and the more that everyone's using AI native coding tools or coding assistants or IDEs, there's really no need for everyone to have a license of Jira or everyone to have a license of any tool when you just plug the AI tool into like an MCP server that's talking to all these tools natively. It can all work as a single service account. There's really no need for multiple seats. So then, we need to monetize our products differently and it's like how much activity is Jira seeing and let's charge for that versus a per-user count and pricing it correctly so we don't lose revenue going forward when we switch from a seat-based to a consumption-based, unless there's some third pricing model that people can come up with that helps keep these companies afloat. I am concerned with all these companies, not necessarily just Atlassian, with, okay, we convert to this type of model to keep revenue flowing, but at some point AI is just going to take this functionality in-house and we won't need AI to manipulate Jira. We'll just do it internally inside the AI tool. And so at that point, why do we need these software tool vendors? I think it's going to be a struggle for them to convince us that they're still relevant. And if AI doesn't go that direction, if they just focus on continuing to improve, get rid of the hallucinations and wrong answers and get even... and we're seeing it go in that direction. It's getting better and better weekly. If they just focus on that and they leave the tool vendors alone, the tool vendors will survive. If we are in a world where the tool vendors are just partnering with AI and they're just getting better and better at input and output via the AI coding tools. Then they can survive if the Anthropics and the OpenAIs decide, yeah, we don't really need you guys. And there's no technical reason why something like a Jira has to exist once AI gets to that point. So they could go through the entire software development lifecycle set of tools, all the DevOps tools and just replace them one by one with internal features to the AI tool.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 反对 AI 采用到达钟形曲线顶点后，开发者会出现陡峭的大规模裁员，时间约 2-3 年
adopted AI in software development and that it's trustworthy and we're all okay with it, as we go down the other side of this bell curve, that's where we're going to see massive layoffs of software developers in these organizations. And that other side of the bell curve may be very steep. The big question is what's the timing on this? I think it's a 2 to 3 year march...
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 反对 AI 工单拦截让公司少招约 4 名支持人员（JSM 按 agent 计 seat，推断 agent seat 增长被压缩）
I think it sits pretty high on there because we actually only added one additional engineer and then we added a manager for our support team, so we segmented things. Instead of hiring, I think we were maybe four to six people was on the docket. We actually hired two people. We saved four headcount for this one product and we haven't actually seen a huge uptick in terms of stress for our support team and ticket count, just because we've been able to deflect so many tickets out of the queue into Console. I'd say well worth it, it's probably even more so because of just the speed at which it answers questions, gives access, things like that. It works 24/7. We don't have to wait for somebody to be on and go through their tickets and finally give some of access. Somebody asked for it, if it's available, it'll do it. Very, very high on the ROI.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户 - 反对 30-40% 的工单经 Console 被拦截，IT 从未看到
Yeah. Basically, we had a 45-day pilot from February into March, and then April we signed the contract and implemented it in production. I think we initially did a 500-person demo or pilot. From there, we opened it up to everybody. At this point in time, everybody uses it. We have probably between 30 % and 40 % of all tickets that get submitted through Console are deflected, and so we never even see those. A lot of the ones that do get sent up to us are more so one-off request or one-off incidents where they need to be reported and handled by an individual or that's like a hardware issue. "Hey, I cracked my screen, I need a new computer." In those cases we need somebody to be involved because we need to figure out, what computer do they need? Do we send them a used one that's older, do we send them a fresh one since maybe they're due for a refresh in the next six months or anyways, things like that that take a little bit more foresight in interactions. I think for the most part, any type of access requests and just general questions, I'd say majority of those are handled by Console at this point.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户 - 支持 CEO：没看到客户有任何 seat 压缩信号，反而相反
Yes, Gregg, thanks for that and for calling it out. I hope we've been very consistent on our views in that world. We are not seeing any signal of seat compression from customers. If anything, we are seeing the opposite. We are seeing strong expansion numbers, strong cross-sell numbers between collections, strong usage of AI and strong commitment to the Atlassian platform. Many, many competitive wins, a huge amount of consolidation happening into the Teamwork Collection. So we have a lot of green lights in a lot of different places. Similarly, you can see that NRR maintained north of 120% and even ticked up again for, I think, the third or fourth quarter in a row. We have a lot of reasons why that is. Firstly, I would go to high R&D investment in just great quality software, right? That does matter. It always gets ignored in a lot of context, wrong choice of words, in a lot of quarters. But we build amazing applications that deliver great value for our customers. Secondly, the context we have in the Teamwork Graph and the critical business processes continue with AI to blur the boundaries between teams and between roles. That means our platform and our offering between service teams on one side that connect to finance and HR, marketing and in through ops teams to technology teams with business teams and leadership teams. The same context across all of those teams allows us to expand into those nontechnical roles at an increasing rate, right, which shows why we are getting that seat expansion in different collections, in different areas and just the continues in our business. Fundamentally, customers are opting for more and more workflows on the Atlassian platform.
- 支持 Jira seat 扩张与 TWC 升级同时贡献 Cloud 超预期，削弱'AI 杀 seat'论
- Mgmt framed the upside as seat expansion plus cross-sell, not a one-off migration beat.- Rovo added millions of MAUs, AI Rovo credit usage grew >20% m/m, Rovo customers grew ARR ~2x non-Rovo customers, and Teamwork Collection is becoming the main AI monetization vehicle through larger embedded credit pools and more agent usage. The print weakens the “AI kills seats” thesis because Jira seat expansion and TWC upgrades both contributed to Cloud outperformance- Mgmt said it is seeing “the opposite” of seat compression, with strong expansion, strong cross-sell, strong AI usage, and NRR north of 120% with another sequential uptick.- Service Collection surpassed $1B ARR, grew >30% y/y, and mgmt called this TEAM’s largest-ever quarter for competitive displacements from a major ITSM provider (neg read-through to NOW)- Data Center Revenue $560.7M, +44.3% y/y vs Street $517.1M, +33.1%, benefited from ~$50M of greater-than-expected upfront term license revenue, helped by EOL dynamics, the large Q3 expiry base, and a March pricing change. Mgmt said Cloud migrations are on track and retention is robust, but Data Center customers preparing to migrate are moderating seat expansion, making FY27 DC deceleration steeper.- Non-GAAP OM was 34.0% vs Street 27.7%, and FY26 OM guide was raised to ~29.0% vs prior ~25.5% and Street ~25.6%. Bulls view this as evidence of Cloud platform leverage, restructuring savings, and expense discipline. Bears note the $50M upfront DC revenue benefit naturally lifts margin and Q4’s restructuring savings should not simply be extrapolated into FY27.- FCF $561M missed Street, with the miss tied to restructuring payments and billing/collection changes, including Data Center contract duration moving toward one year and annual billing on some multi-year Cloud contracts. Billings were also soft versus Street, but mgmt is trying to steer investors toward subscription ARR disclosures at the upcoming investor day to normalize revenue-recognition noise.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 反对 空头：AI agent 是否最终降低 seat 型协作/规划软件需求仍未解决，尤其开发者重度场景
Bears will also argue the AI debate is not over. Rovo traction is encouraging, but the broader question remains whether AI agents eventually reduce the need for seat-based collaboration and planning software, particularly in developer-heavy use cases. TEAM is also facing more competition: Service Collection success is encouraging, but ITSM has a very large incumbent in NOW, and AI agent platforms from hyperscalers, CRM vendors, dev tools, and foundation model companies all want to own workflow context. The FCF miss and billings miss give bears additional ammunition that revenue-recognition transitions, annual billing changes, restructuring payments, and DC-to-Cloud migration mechanics can keep the model noisy for several quarters.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 支持 CEO 认为 AI 时代开发者会更多，agent 会带来更多需管理的流程，利好 Jira 这类编排平台
Now you might be asking, okay, they're adding seats, but seats for whom, right? And this is where the old narrative, I think, breaks down. People pigeonhole Atlassian as a developer company. I might argue there's going to be more developers in the AI era. I think that's going to be true. That's fine. You think we're a developer company, or some people do. We love that heritage, I will say.
- 支持 agent 数量激增（100 人公司可能有数千 agent）带来更多管理开销，利好 Jira；但这是对流程量而非人数 seats 的论证
It continues to grow really strongly and we ask a lot to make sure that it is the case. At the same time, in the Agentic world, what has been done, who is doing what, to which piece of any workflow is -- needs to be tracked, managed and understood. I think if you want to think about it generically, I believe you'll have way more actors doing that, right? The idea that a 100-person company might have thousands of agents running around, that creates a lot more management overhead and understanding overhead, which I think is really good for orchestration platforms like a Jira that already have workflow and automation and all sorts of other capabilities to handle or manner of business workflows and processes. So I talked on a podcast and someone sort of -- I think one of the hosts explained to me backwards or I figure which way it went, but you're sort of talking about system of record and system of record. And that always sounds to me like a very static database-driven type of thing when you're a system of record. We were sort of -- maybe not even accused or pointed to being, I forget how it happened, to a system of process.
- 口径不齐 团队开始让 AI agents 与人协作被作为 seats 继续增长的理由之一，但未说明 agent 是否算 seat 或收费
Why? Because workflows continue to get more complex, more intertwined, Teams and companies are buying more into the Atlassian platform and getting exponential value as they increase their commitments. And that's really important, and we think that only keeps accelerating from here. The teams start orchestrating the AI agents alongside their people. You heard me talk a little bit this morning about the Teamwork Graph. I think we have more ahead of us that we can continue to do because of our investments in that platform. Brian? Now you might be asking, okay, they're adding seats, but seats for whom, right? And this is where the old narrative, I think, breaks down. People pigeonhole Atlassian as a developer company. I might argue there's going to be more developers in the AI era.
- 口径不齐 CEO 称工程不再是瓶颈——是论证 PM/反馈环节的新机会，但也隐含 AI 提升工程效率（对开发者 seat 数的含义未展开）
At the same time, it leverages our strengths in AI because I do think the volume of feedback is going to go up, the product manager's job to take all of that feedback and distill it down and turn it into ideas that become the pipeline for engineering, if you want to think about it that way, you've got to first start with a set of insights and sort of scraps of text and other things from customers. You distill that into a set of ideas. Those ideas become road map items and road map items get built by engineering. Engineering is, as ceasing to be the bottleneck in that process. So that creates other economic opportunities when you go outside of that. Now a product manager job isn't going to be as classical. It's going to have a lot of AI and automation and other things associated with it, which is exactly what we're trying to do in the feedback I to show some of this at a large scale in a category where we don't have much footprint. So I think it's all upside for us, which is growing really well.
- 支持 Rovo 客户 ARR 增速为未开 Rovo 客户的 2 倍；MCP 用户 ARR 增速也为 2 倍（相关性，非因果）
So as we said in our shareholder letter, Rovo customers' AI credit usage, their rovo credit usage is growing at 20% month-over-month. That is a pretty good number. We're early in the curve, but we're already a very large player in terms of AI MAU and usage and the density of usage is growing very fast. Secondly, when customers adopt Rovo, they grow their ARR at twice the rate -- twice the growth rate and our expansion rate, as you know, is quite good, our NRR. They're growing at twice the growth rate of customers that haven't turned on Rovo. And lastly, it is delivering value, right? They are stickier. They move more. Again, we've given all these stats. I don't know if the 75% of the Fortune 500 companies that have turned on Rovo is a new stat, but it's there, and you can see our Agentic automations are growing really, really strongly. One way of using agents, again, is through automation, putting them into various rules that happen automatically and can access all of the Teamwork Graph.
- 口径不齐 CFO 提到企业内 AI 使用不均：有重度用户，也有已开通但不使用者——credit 池化可在组织内调配
And Mike touched on this a little bit earlier. But just maybe to put a finer point on it, TWC is the best AI monetization vehicle right now. And a big part of that is because of the credits that we're giving 10x credits relative to the stand-alone SKUs. But I think what we hear from customers is twofold. One is they want the access. Two, they also have within their enterprises, and I'm sure you guys see this within your organizations as well. There's lumpy usage. You got some hyper users, others that are provisioned but not using it. So they're able to load balance those credits across their organization as they're getting better and better at driving value from that. So it's giving them that access, it's giving them that predictability from an invoicing perspective. Then Mike shared a little bit earlier, all the different commercial models that we will adapt to as our customers are evolving as well. So I think we're seeing that. One of the things I don't want to get lost here is when we showed the video a little bit earlier there. I'm not sure if we ended up showing the stat, but in the keynote, we showed a stat that talked about how we were getting 44% of better quality in terms of output and 48% cheaper rate, right? You're using 48% fewer tokens to get to that same or better outcome. That's a conversation with your CIO. That's a conversation with your CFO that's going to resonate. And so I think both things are true. We're seeing that show up today in our monetization, and we're still building the platform in the flywheel to continue to compound on that.
- 口径不齐 QA 团队缩小但工作量扩大，seat 持有人转向开发/产品负责人（人数不一定增）
**Expert:** Fewer manual testers, that's a trend, and that's been going on even pre-AI as well as developers are taking on more ownership of quality. Although the QA and testing group has been shrinking at most enterprises, their work is expanding because there's still a lot of really poor quality software getting shipped, and I think AI is amplifying that. I don't think companies are going to come back and hire any more humans to be on the QA team. They're pushing it back into the developer and product owner team. There's still people who have accountability to the quality of their software. It's just the seat owners of some of these tools might not have QA or testing in their title, but they're accountable for quality. I think a lot of it is kind of shifting back into the developer. That trend, again, has been going on before AI took over. I think there's an integrated product engineer owner type title in the new world of AI or an AI agent manager title. There's new roles that are forming as we speak that will own the quality of what software gets shipped that you can't really outsource to an AI, but the titles are definitely changing. The roles of who owns the development cycle are changing. They'll look different in three months from now, six months from now, but there are people who own the performance of the software and there are customers who are going to complain when something doesn't work right.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 支持 工作量回升、新角色增多后，收入对人数的敏感度会下降
**Expert:** SmartBear's largest customers who were going through just workforce reductions, there's definitely dollar-based churn when licences were going down. There was certainly sensitivity on GRR level if customer lays off 1,000 people and 100 are developers. You just don't have as many seat licences to sell. That was not the majority of customers that was impacting several of large enterprises and a bit of downward pressure on gross revenue retention as they just don't renew with as many licences. **LF:** Is SmartBear's revenue becoming less sensitive to who is doing the work, and more sensitive to the volume of work? **Expert:** Yes, and I think that then becomes the pivot of moving away from just that seat licence is now the volume of work is going back up again, even the number of developer or just people with the new titles, but they're focused on delivering software is increasing and shifting that churn sensitivity goes down. It becomes more stable.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 反对 seat 型模式风险：若 AI 让 2 人做 5 人的活，厂商损失 3 个 seat
RG: I've mentioned it a couple of times, but I still think the biggest risk is just the behemoth companies like the Microsoft and Google. People have almost begrudgingly bought Asana because their existing productivity provider hasn't had a good solution. I don't know if anyone listening to this call has ever played with Google Tasks or Microsoft Tasks. It is very basic. It is a to-do list at most because these companies have had bigger fish to fry. But as soon as Microsoft and Google decide to put any effort into their Work Management or Project Management functionality, it's so attractive for a company who's already using Google or Microsoft services to just expand their usage to that new project management or work management functionality. There will be synergies and benefits to having that live within their existing tools and tech stack, and then there's going to be humongous commercial incentives as well. They'll probably make it super cheap, even free. That's the bundling effect where you're paying for your email and it looks like Teams or the project management is free. So I would say Microsoft and Google are still the threat. Another risk I would see is what I was talking about before, so I won't rehash it, but just about becoming a glorified database. You could ask Claude to make and complete tasks. You can ask Claude to analyze your work and you frankly start to not even spend any time in Asana. It becomes less and less valuable. And then if the whole AI dream does come to fruition and people become less needed, well, that hurts Asana's business model because Asana is not based on outcomes. Asana is based on seats. So if one company only needs 2 people to do what they previously needed 5 people to do, well, Asana has just lost 3 seats, so that's a risk as well.
来源背景：Robbie Goldberg，前 Asana APAC Corporate Sales 负责人 - 反对 用户可以让 Claude 创建/完成任务，工具沦为'数据库'，在工具内时间减少
RG: I've mentioned it a couple of times, but I still think the biggest risk is just the behemoth companies like the Microsoft and Google. People have almost begrudgingly bought Asana because their existing productivity provider hasn't had a good solution. I don't know if anyone listening to this call has ever played with Google Tasks or Microsoft Tasks. It is very basic. It is a to-do list at most because these companies have had bigger fish to fry. But as soon as Microsoft and Google decide to put any effort into their Work Management or Project Management functionality, it's so attractive for a company who's already using Google or Microsoft services to just expand their usage to that new project management or work management functionality. There will be synergies and benefits to having that live within their existing tools and tech stack, and then there's going to be humongous commercial incentives as well. They'll probably make it super cheap, even free. That's the bundling effect where you're paying for your email and it looks like Teams or the project management is free. So I would say Microsoft and Google are still the threat. Another risk I would see is what I was talking about before, so I won't rehash it, but just about becoming a glorified database. You could ask Claude to make and complete tasks. You can ask Claude to analyze your work and you frankly start to not even spend any time in Asana. It becomes less and less valuable. And then if the whole AI dream does come to fruition and people become less needed, well, that hurts Asana's business model because Asana is not based on outcomes. Asana is based on seats. So if one company only needs 2 people to do what they previously needed 5 people to do, well, Asana has just lost 3 seats, so that's a risk as well.
来源背景：Robbie Goldberg，前 Asana APAC Corporate Sales 负责人 - 反对 若 agent 做 20 人的工作，就不需要 20 人；纯 per-seat 厂商必须转用量否则消亡
JS: Yeah, for sure. So a good way to think about that is, you know, we may be seeing something like a transition right now. Who knows if it will always be a seats plus credits. It might one day just be credits like usage. And especially if AI agents can do the work of 20 humans, why do you need 20 humans? You can just have one powerful person that is, you know, a big operator and then pay for usage instead. So it's a very interesting time where companies that were existing only on a per-seat model have to somehow transition into a usage model or they'll die. So a good way to
来源背景：Jessica Saranich，前 monday.com Regional PMO Director & Chief of Staff - 支持 AI 降低编码门槛，非开发团队也在产出代码，推高规划/追踪/治理需求
On the AI side, it's reducing the barrier to entry as it relates to coding, right? So you're going to see a proliferation of code. We're already seeing that coming out of developers. You're seeing a proliferation of code coming out of non-developer teams as well, marketers, legal, finance, et cetera. And you're seeing more innovation, more software, more applications being built day-to-day. And we see that in the hundreds of millions of workflows that flow through our platform. And so that's doing two things. One, it's creating a lot of proliferation as it relates to the need for planning, tracking, governance and collaboration. And that's exactly where Atlassian's platform sits. The other piece is then that not all AI is created equal, right? When you think about the agents and AI that is much more intelligent, it's going to need context. And I think we're seeing and hearing more and more of that right now. So when I think about Atlassian's 20-year history and the investments that we've made, not just in applications, but more and more on the Teamwork Graph that we've talked about is that we're able to deliver higher quality outcomes and outputs through that Teamwork Graph for our customers at a much better rate as well. So those are some of the things that really stood out to me, Brent.
- 反对 开发者会变少，预算流向大模型厂商，客户会主动削减用户数
**Expert:** I honestly cannot think of a world where it wouldn't work. I think that you can prove it with… I think the consumption-based model is the only one that's going to work for any company that's in SaaS at this point, in my opinion, because naturally, you're going to have less developers doing things. You need less people to do things now with the tools at hand. Most of the budget right now is going towards the large LLM models, Anthropic and OpenAI. Really, when you move all that money of a company to those purchases, then you have less money to spend on seats on GitLab or any other tools. Company is naturally going to figure out ways to cut down the number of users or if you could pivot to consumption model, you can actually keep those contracts steady and you don't have to churn in the next year or two at renewal. The consumption model is basically a way to financially… Accounting-wise, you are financially redoing the contract so that the customer doesn't feel like they're not getting the value add. Because if they lose users and you're still on user pricing, you can't convince them to pay 50% more per user. That's just not going to work. The cost of switching now is so much easier than before. The customer is always going to win that negotiation. If you move to consumption, it's like you can find a different way of modelling the next contract at renewal of like, "Hey, yes, you have less users, but here's how much more you're using. This is why this new pricing model works for you." It's easier to justify to a CFO who's trying to sign the contract. **LF:** Would you expect GitLab to price by tokens, tasks or some other measure within the new pricing model? **Expert:** It's going to be tokens. I think they haven't released what their strategy is, but it sounds like it's going to be tokens and very much like Claude and Cursor, all those other companies.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 反对 用 Cursor/Claude Code 生成代码后，客户质疑是否需要这么多 seat
**Expert:** The tier down's biggest thing for GitLab specifically, it's the most expensive product is a security offering that goes on top of your normal standard platform. That security offering is no longer a value add anymore. That could cause some churn and tier down. Then, you have the web IDEs like Cursor, Claude Code doing all that work every day anyways. All you're doing is just prompting to create code into GitLab. You start questioning like, "Do we really need to pay for this many seats when you just have one developer doing all that by themself?" You can only pay for one developer seat of GitLab. I think those are the biggest risks. Then, you have other risks as just I'd say, maybe it's easier now to migrate off of GitLab or GitHub. It used to be very, very sticky. When I was there, any time we have a renewal, it was very easy to negotiate a renewal because you can literally just say, "Hey, you're already using our CI/CD, you can move off of GitLab, but it's going to take you two years to do it. It makes no sense." The cost of switching was so high, you can upcharge them 50% more on renewal and they wouldn't even want to move. They wouldn't even consider it to just take the 50% increase. Now, because of AI, it's actually pretty easy to migrate off of GitLab or everything is just easy to migrate in general. You really have to be prepared to win that customer and renew that customer without the threat of like, "Oh, it's just too sticky," because nothing is really sticky anymore. It's easy for anyone to just move off of anything these days.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 口径不齐 未来人类用户被 agent 取代，但仍需少数人管理 agent 的平台（对‘管理层’类产品是机会）
**Expert:** Problem is it's the pricing model, I don't know how many people are going to have in 2-3 years, "Why we do we see base pricing?" Every company is revaluing how many people do they need for the next 2-3 years in developing it. That's number one. I think number two is I think if they figure out the pricing model, and then, they figure out how they can be the need to have in the agentic era, which I think they're trying to do, if they're trying to get into a company and they're like, "We could just use GitLab for Git. We don't need anything else. Can you just give me a very watered down version of GitLab?" I think if they can sell the fact that, "It's like all or nothing." You need the platform. Then, we're going to have a way to manage all your agents on this platform. If they could figure out a way actually to have a context layer on top of GitLab like a portal layer where they can manage all these agents within GitLab, I think that's really the biggest opportunity because in the end, it's like you're replacing all the human users with agents and agents are working 24/7, but you still need someone to manage all those agents as a human being. If you could build a platform that can allow two or three people in that company to orchestrate all the agents and understand what work they're doing, how much code they are pushing to production, you can track everything metrics-wise, that's going to be the biggest opportunity.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 支持 针对 seats bear case：AI 降低编码门槛→更多软件和协作；买 TWC 的同时 Jira 独立 seats 继续扩张
On the bear side, definitely a discussion on seats. What does AI do to seats over the long term? And so why or why not is this a worry for Atlassian, whether that's -- does AI create more seats, pricing model changing? Like how do you -- how should we think about as AI and seats not a bear factor for Atlassian? Yes. So I think the first thing is we really take a long-term view of this, right? And if you believe that AI is going to lower the barrier to entry as it relates to coding and as a second order effect create more innovation, more software, and as a result of that the need for more collaboration and workflows will be created, that's exactly where Atlassian sits. And even today, within our results, as I mentioned a little bit earlier, as customers are continuing to adopt more AI through purchasing of Teamwork Collection, we're continuing to see stand-alone Jira seats expand, right? So I think this is really customers voting and saying, look, Atlassian is a long-term partner for us. As it relates to our own AI strategy, it's going to be a key platform that we continue to invest in. And that's really the thesis as I think about like the long-term opportunity for us. We're seeing it in the results today, and we believe that it's going to be a tailwind for us through our existing platform.
- 支持 AI 降低编码门槛，创新扩展到业务团队，需要更多协作/规划/追踪
And AI is an accelerant on top of that. And so when we think about all of that coming together, it becomes an incredible moat for Atlassian. The other area is the orchestration, right? So if you believe that AI is going to drive down the barrier to entry, drive down the barrier to entry for coding, that's going to create more innovation, more software, more applications, not just in engineers and developing teams, but across business teams as well. We're seeing that across our customer set already. As a result of that, more collaboration is needed, more planning, more tracking, the governance, the permissions, the administrative ability to think and look across the -- all the different workflows and applications that are happening right now, all of that already exists and sits within Atlassian's platform. And we've got 20 years of history here, connecting technology teams to business teams and applications. And so all of that helps build that defensible moat.
- 支持 MCP 用户增速 2x，Teamwork Graph 写入对象月增 100%，MCP 客户 MRR 增速 2x
Yes. I mean I think, again, you can see it in the performance already, right? We talked about -- we introduced ARR at the Investor Forum growing north of 20%. And if you look at the NRR on our cloud basis as well, north of 120%. So incredibly healthy. We called out the outperformance in Q3 due to Teamwork Collection as well as continued growth and expansion in Jira stand-alone seats. And on the AI side, a couple of the areas that we had shared was the fact that our Teamwork Collection customers, they're using Rovo credits at 2x the rate and deploying 2x the number of agents as those that aren't yet deploying Rovo. Our Rovo customers are growing and using their AI credits 20% month-over- month. And Koji, you asked a question about maybe some stats that we haven't necessarily emphasized as much. If you think about the platform, I talk a lot about Teamwork Collection and then multiple platforms as well, but again, open by nature, right? There's customers certainly that use their own agents MCP into the Atlassian platform. And that, again, compounds back to the Teamwork Graph, and we're seeing MCP users grow at 2x the rate. The number of objects being written to Teamwork Graph right now is growing 100% month-over-month. This is through via MCP, right? So as customers continue to develop their agents, as customers continue to use Rovo agents on top of our platform out of the box, we're seeing incredible adoption across the AI sector. And I would also mention that those customers that are using MCP, their MRR, it's early days, but MRR is growing 2x the rate. And these aren't just existing customers, these are customers that have been both with us for a very long time and also what I would consider as AI native customers, right? If you think about the Forbes AI 50, 2/3 of them are Atlassian customers.
- 支持 AI 提高速度和吞吐，人与 agent 之间需要协调，需要更多人加入平台
Yes. I think certainly, Q3 numbers kind of back up a lot of what we certainly have believed. And I think when you think about the complexity of the world and the speed in which AI evolves things, that -- the need for collaboration, coordination raises more than ever, right? Like you need to coordinate more across what are we doing across the organization, especially as the speed and throughput increases and certainly, not just coordinating between humans, but also coordinating between humans and agents. Are we doing the right things? Are those agents actually delivering what we want them to? If Gregg and I are on the same team, we're making sure that as we deploy agents, we're not overlapping, having kind of conflicting things that we're working on. Are we actually progressing against our strategic results as a company. And so that coordination aspect becomes more important than ever. I think we see more people needing to be added to the Atlassian platform to coordinate that. And then you see things like our innovation of adding or being able to manage and orchestrate agents in Jira. That's a prime example. Jira today is already the orchestration workflow platform for work. Right now, I can assign work to my colleagues and team, and now I can assign work to agents directly in Jira and then the agents can take action directly from there. And again, so managing what are our teams working on, managing now, what are those agents working on? Are they delivering actual tangible results for us and advancing our strategic priorities as a company because that's actually what matters, right, and not just the volume of work.
- 支持 Rovo 活跃客户 ARR 增速约 2x，带来更多使用、更多用户
Yes. I don't think anything has changed technologically. I think it's the steady progress that we make across kind of all of our 3 priorities and especially in advancing Rovo. When I think about Rovo, very early on when we introduced it several years ago, it was centered around enterprise search. And then, of course, then chat functionality. And now there's more agentic usage. And that, of course, is driving the increased Rovo credit usage. You're seeing Rovo credit consumption grow 20% month-over-month, which is incredibly fast and really good to see. We're also seeing, interestingly, those customers that actively use Rovo versus those that don't actively use it or don't light it up, grow their ARR at about 2x the rate of those that don't. So that's a good proof point. I think that when customers utilize Rovo, they get incremental value that delivers more workflows through the Atlassian platform and that then leads over time to more usage, more users and then more monetization opportunities down the line.
- 反对 AI agent 在 Notion 内基于完整上下文生成工单，使 Jira 不再必要
SG: Okay, but basically we have a backlog for me, as far as I see it or I understand it, for me it's just some database. There is a database of tickets with different properties. For instance, we have also a database of feature teams. So every ticket will be related to 1 single feature team. So you have a database of tickets, a database of feature teams, you have a database of sprints, you have a database of release, you have a database of epics. And if you know how a backlog is working, you can connect all of these databases with exactly the same things and the same capabilities as in Jira. But the main advantage is that you have your backlog inside your whole documentation and your whole knowledge. So in terms of context, in terms of... For instance, we have also a database of PRD, product requirement documents inside Notion. So it's very easy to then create... what I have also built... an agent to create the tickets for the team. So the product manager, for instance, queries this agent. The agent has the whole knowledge of this feature team and the different projects they are working on. And the most important thing is the context and the access to this context. And if you have everything inside Notion, you can pretty much do... It's very powerful. You don't have to connect to Jira, you don't have to connect to Teams, you just have everything on Notion, and that's for me the... The way to create a very strong backlog and the main advantage is that it's free. The additional cost of creating a backlog is free, it's not more expensive, you can cut out your budget because you replace the whole Atlassian suite with Notion directly. So for me, it's very interesting.
来源背景：Sidney Golstein，Rakuten France Lead AI PM；Rakuten France 约 200 人 - 支持 AI 让客户做更多原本做不到的工作，因此 seat 数没有像预期那样下降，反而上升
**BR** [00:02:06]: I've been selling AI as part of workflow products for the better part of 10 years, the first iterations of AI, of course, being machine learning-based, and then that's changed over the last years to agentic and generative AI capabilities. During that time, what I found is it's hugely valuable to customers and to those workflow solutions because, really, what customers are looking to do is have their problem solved, and AI can come in and actually make workflow solutions more efficient and effective. I'd say one other thing, and this will come up probably in some of the other discussion points today, what surprised me as I've been selling the AI software to customers, in the entire time, I thought the seat counts were going to go down. While I was at ServiceNow, in particular, when we were talking, we expected that seat counts would go down as people bought more AI. The reality was the opposite, and the reason that it was the opposite is because, almost all of the customers that we were selling to in Atlassian is similar in this regard, all of them are barely touching the work that they should be doing anyway. At most, prior to AI, they were getting to about 10% of the potential work that they had on their plates. AI is helping them get to, what, 30%, 40% of the work now. There's still a lot of untouched work, so it doesn't mean that seat counts go down. It certainly means that workflow software can be further made more efficient than it is and, therefore, really a benefit, to your point, to Atlassian and to other SaaS companies as well.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 AI 前客户只完成约 10% 的潜在工作，AI 后 30-40%，仍有大量未触及工作，所以 seat 不会下降
While I was at ServiceNow, in particular, when we were talking, we expected that seat counts would go down as people bought more AI. The reality was the opposite, and the reason that it was the opposite is because, almost all of the customers that we were selling to in Atlassian is similar in this regard, all of them are barely touching the work that they should be doing anyway. At most, prior to AI, they were getting to about 10% of the potential work that they had on their plates. AI is helping them get to, what, 30%, 40% of the work now. There's still a lot of untouched work, so it doesn't mean that seat counts go down. It certainly means that workflow software can be further made more efficient than it is and, therefore, really a benefit, to your point, to Atlassian and to other SaaS companies as well.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 AI 创造工作：开发产出翻倍后需要更多销售、支持人员（数字为专家自称编造的示例）
**BR** [00:05:03]: Certainly, it's my opinion, but I don't see revenues and seat counts necessarily going down, and part of the reason is the… I don't know if it's a secret or not, but AI creates work. The more efficient and effective you get… Let's just take an example which fits right in Atlassian's area. Developers, let's say they used to be able to develop two applications a quarter. This team used to do two applications a quarter. With AI, they can double their output, again, just making these numbers up. They can double their output. Who's going to support those four applications now? It used to be the same team supporting two applications. Now, they're supporting four applications, all of which… They're now being sold to customers. It means you have to have more salespeople actually selling to more customers. You have to have more support people actually supporting more customers. The same development team has to be more effective and efficient to be able to support now four applications as opposed to two.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 专家认为这是持久而非暂时现象
**BR** [00:07:40]: Yes, that's exactly right. I've got a lot of examples. Again, I've been selling this for so long, and I'm just taking the IT service management space as an example. We've had AI there for a while. It's been able to deflect a lot of calls that were coming in. People who were calling for a password reset now can just be handled through a self-service AI interface, but what's happening to the labour force on the back end is that they're having to adjust. Rather than having these simple requests that are coming in, they're actually now having to take more complex issues and solve them because AI has got all the easy stuff. I think you'll just find that, in every sector and everywhere where AI is, human and AI together are going to be able to solve more complex problems than just a human alone would have and also more than just AI itself would be able to.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 CEO 判断 5 年后开发者更多，客户信号显示开发者招聘会继续增长
Look, I can talk about it from a developer or engineering front in a number of different angles, Gregg. I know we've talked about this before. We've been long on record. You can look at the receipts I think there's getting more developers in the world in 5 years' time than there are today. The cost of building technology is going down. The amount of technology we're going to build is going to go up. I believe that will continue to be a good trend. We've said this for a number of years. Those are the signals we see in our customers, the ability to grow that is huge. I think secondly, the number of people building technology that we don't necessarily call developers is also growing. So we're seeing people in finance and in HR and in marketing, using a lot of these tools to build technology of various different forms. Again, we see that in our MCP server usage and our CLI usage coming across the board in lots of different functional areas, not just inside the engineering fraternity. And I think that's really important. From our perspective, again, the system of work is about helping companies to connect their technology teams with their business teams to have a single platform that allows both sides of the organization in a technology-driven organization, an organization where software is their key competitive advantage. It doesn't make them a technology company. It can be a bank, an insurance company, a health care company, a university. There are lots of companies for whom. Many organizations in the world, technology is a key competitive advantage. That is -- I believe that is a big growth cycle that we are continuing to participate in. Our strength is in connecting the technology team through the business team and I do believe that the developer hiring will continue to grow as we see in our customer signals.
- 支持 管理层认为部署更多 agent 增加协作/规划/协调需求而非消灭 seat；开发者与非开发者构建软件的人数会增加
Teamwork Graph now spans more than 200B objects and connections, with TEAM claiming up to 44% more accurate answers using 48% fewer tokens. Mgmt expects enterprises to maintain three to five major knowledge graphs and intends for TEAM to be one of them, while opening the Graph to third-party agents through MCP and CLI rather than trying to own the entire AI stack. * **Demand** Mgmt described customer demand signals as healthy and broad-based, with continued expansion in both technical and knowledge-worker seats. Roughly two-thirds of Jira and Confluence users are knowledge workers, and mgmt argued that deploying more AI agents increases the need for collaboration, planning, workflow management and coordination rather than eliminating TEAM seats. Mgmt also continues to expect the number of developers and non-developers building software to increase. * **Margins** F4Q non-GAAP OM of 36% beat Street by roughly 540 bps, and TEAM reported its first GAAP-profitable quarter with a 12% GAAP OM. FY27 non-GAAP OM of 25% looks weak against Street at 28.8%, but includes roughly four points from lapping FY26 Data Center revenue-recognition benefits and another three-point headwind from moving compensation from SBC toward cash. Adjusting for both, underlying non-GAAP OM is expected to expand roughly 150-160 bps.
来源背景：TMTB 作者转述业绩会与 callback - 中性 分析师质疑开发者/知识工作者招聘回暖是否是短期现象，离岸+AI 可能带来收缩；IR 定性回应 AI 让更多公司写软件
KeyBanc Capital Markets Inc., Research Division Yes. That's a good reminder. As, like, a software analyst, that's new to me. New concept. So maybe if we go back to kind of your explanation on the paid seat expansion. So one thing that investors have been focusing a lot with a lot of the AI worries has been like developer growth and developer headcount and knowledge worker growth and knowledge worker headcount. If we look at different data sources like Indeed job data, we do a CIO survey, and hiring intentions are up. There's a number of other data sources that are also pointing to near-term positive indicators for developers and knowledge workers. What do you think is really catalyzing this? Do you think AI is catalyzing near-term activity? And that this is maybe just a flash in the pan before we eventually see some contraction. We have some members of our KeyBanc IT organization here, and they talked about labor arbitrage with offshore and AI potentially. So help me understand maybe what you think around this.
- 支持 AI 降低软件开发成本，非科技公司也成为软件公司，需要团队协调
Yes. I think it's important to note, as I mentioned earlier, that, that strong seat expansion or strong seat growth that we saw in Jira and Confluence is actually across both software development and importantly, non-software development or knowledge workers. So teams like HR, marketing, legal, finance. So really good traction there, and that continues to be an area of focus as we focus on penetrating more of the enterprise customers that we have and reaching and serving more of those users. But on the software development side, I think it's an indication of ability to create software now with AI is becoming greater than ever. And you're seeing that ability for companies to become software companies, whether you're a traditional nontech company, you now -- all of these companies have to drive more digital transformation, create more software, create more digital services as part of their overall strategy. And AI is lowering that cost, you're able to drive a lot more software development. Now the next part of that challenge is how do you actually make this -- all this increased software move in the direction for value for your customers or for the enterprise customers. And that requires a different level of coordination, right? Like a lot of the AI capabilities we've seen to date have been focused on personal productivity or individual productivity. Our ticker symbol is TEAM. We've always been focused on teamwork. How do you, again, I mentioned earlier, coordinate across your organization. How do you manage, track and plan work to make sure that we're all moving in unison towards the organization's strategic goals and delivering value to your customers because that's actually what ultimately matters as opposed to, again, all these disparate individual tracks happening. And I think that highlights what you're seeing in play.
- 支持 第三方 agent 经 MCP 接入：MCP/CLI 月活超 100 万，经 MCP 创建的 Jira 工作项和 Confluence 页面 4x，MCP 客户 ARR 增速 2x
That's via increased stickiness, faster seat expansion and higher up-tiering to higher value editions. Actually, what's also really interesting is we shared a bunch of MCP stats. So even if customers are using third-party agents, and tapping into the Atlassian platform via MCP, we're seeing significant growth there. So we have over 1 million monthly active users of our MCP server and Teamwork Graph CLI. So again, wanting to tap into the value of the Teamwork Graph, even if you're using a third-party agent. That's delivering significantly more value to the customer. You're seeing that in their output. So they're driving 4x the amount of Jira issued or Jira Work items created via MCP server and Confluence pages created 4x the amount via MCP server. So they're driving significantly more workflows through the Atlassian platform, which is what we want to see. And ultimately, that increased value to customers comes back to us. So exact same number.
- 支持 Rovo 客户 ARR 增速 2x，回报体现为更快的 seat 扩张
But again, I think the primary motion you're going to see is customers choosing to adopt Teamwork Collections. So I would continue to pay attention to that because that's probably the primary monetization motion today. Customers are also -- when they adopt Rovo, they, of course, realize significantly more value. And we're seeing Rovo customers grow their ARR at 2x the rate of those that don't adopt Rovo. So actually the concept of delivering value first and foremost to customers, then we'll recoup that value back over time. That's via increased stickiness, faster seat expansion and higher up-tiering to higher value editions. Actually, what's also really interesting is we shared a bunch of MCP stats. So even if customers are using third-party agents, and tapping into the Atlassian platform via MCP, we're seeing significant growth there. So we have over 1 million monthly active users of our MCP server and Teamwork Graph CLI.
- 支持 AI 编程工具早期用户 Jira 任务多约 5%、MAU 更高、seat 扩张更快
RBC argues Atlassian can sustain 20%+ subscription ARR growth as AI increases software creation and coordination complexity, making Jira/Confluence more valuable rather than less. The key differentiator is the Teamwork Graph, which spans 200B+ objects and can act as a control plane for agents; RBC cites early users of AI coding tools showing ~5% more Jira tasks, higher MAU and faster seat expansion. Collections/Rovo adoption and improving enterprise GTM provide additional upside, while the end of the multi-year platform R&D cycle should support meaningful margin expansion.
来源背景：RBC 卖方（TMTB 转述） - 支持 coding agent 直接接入 Jira/Confluence 被视为优势；不接入的 Devin 造成两套工单，Jira 仍是工单中心
JL: Yeah, I mean, I would say they're approaching parity. Devin certainly has, I think, a lot more excitement around it, but in terms of its true functionality, it can act asynchronously, it can be developer supportive as well. As I mentioned, it's got dual modality capabilities. And so that I would say is not a differentiator, but it's a superior capability relative to dual modality. I would say it's deficient because it does not have a Shield comparable solution. And so what that ends up meaning is in the Devin environment, we're burning through considerably more credits or tokens than we do in Factory AI because we don't have the same level of observability. And so we're unable to monitor token consumption with the same level of efficacy. And then I would also say that there is some differentiation in Factory as it directly plugs into Jira and Confluence, whereas Devin does not. It's got its own sort of embedded ticketing functionality, which creates, you know, a bit of a lag as we manage essentially 2 buckets of tickets and issues that need resolution.
来源背景：Julian Lustig，Co-operators VP Digital & Data Platforms / Chief AI Officer - 反对 长期看 coding agent 带来减少工程师、架构师、科学家人数的机会
JL: But I mean, you know, your question is, we will never get to 0 developers. I think the value or the vision is that we can either, well, certainly over the next 24 months, it would be to create capacity amongst our developers to tackle more complex, more sophisticated application development, coding, testing capabilities, machine learning development. And utilize either Cognition or Factory to support the lower risk, lower complexity types of activities. Over a longer term horizon, we would expect that both factories' ability to support more complex development improves. And then that creates an opportunity for us to decrease engineer, architect, scientist headcount.
来源背景：Julian Lustig，Co-operators VP Digital & Data Platforms / Chief AI Officer - 中性 未来 24 个月是释放开发者产能做更复杂工作，而非减员
JL: But I mean, you know, your question is, we will never get to 0 developers. I think the value or the vision is that we can either, well, certainly over the next 24 months, it would be to create capacity amongst our developers to tackle more complex, more sophisticated application development, coding, testing capabilities, machine learning development. And utilize either Cognition or Factory to support the lower risk, lower complexity types of activities. Over a longer term horizon, we would expect that both factories' ability to support more complex development improves. And then that creates an opportunity for us to decrease engineer, architect, scientist headcount.
来源背景：Julian Lustig，Co-operators VP Digital & Data Platforms / Chief AI Officer - 支持 编码成本下降→更多软件→开发者和非开发者 seats 都在增加（回应开发者人数担忧）
So, you know, I think as you see the intelligence in the models continue to grow and make progress and compound over time and also seeing how competitive it's becoming from a pricing perspective that's going to benefit the market. There's more value being driven to the market, to our customers. That's going to benefit Osseon as well. And in terms of the C-PACE growth, what that really means is the cost of coding, cost of building is coming down. And so we're seeing more innovation in our customers. We're seeing more software being built, we're seeing more seats being added through across developers as well as Non-purpose. I think one of the things that is most underappreciated about Atlassian is how diversified the set of businesses. You know, I came into Atlassian sort of thinking that it was a toolbase company simply for software engineers called Jira. And the reality is 65% of Juren users are knowledge work on software.
- 支持 '自己 vibe code 一个 Jira' 没有出现在业绩和 seat 扩张中
I think enterprises, organizations that are realizing how incredibly challenging and hard it is to be able to drive A.I. value throughout the enterprise. Right. It's, you know, the days of talking about vibe, coding your own CRM, your own JIRA, etcetera. I'm not hearing that. We're not seeing that in our results. We're not seeing that in our seat expansion the adoption across our collections as well. But really as companies think about where to drive value, it is in that graph, it is in that teamwork graph that we have and that's what's actually opening up a lot of the discussions with the C-suite right now and driving a lot of the results. Thank you so much, James. It's been a very eventful five months, so we're looking forward to see what the next year.
- 口径不齐 AI 没有让研发人数缩减，但工程团队基本持平、低个位数增长，产出增长快得多
从我个人了解到的情况来看，实际情况介于两者之间。AI 提效并没有导致研发人数（developer headcount）同比例缩减；更贴切的描述是工程团队人数保持相对平稳（relatively flat），处于低个位数增长（low single-digit growth），而产出增长得快得多。需要明确的是，生产力提升（productivity gain）并不会 1:1 转化为端到端交付提升（end-to-end delivery gain）。但对公司而言，AI 编程工具确实让单个开发者产出了更多代码、PR 和变更。很多代码现在都是 AI 写的，占比可达 60%、70%、80% 甚至 90%，但这反而极大地增加了下游工作量。公司现在面临更多的 code review、CI/CD 运行、安全扫描（security scanning）、测试、文档、工单（tickets）、部署以及可观测性需求。到了2026年，我认为大约 80% 到 90% 的代码都由 AI 生成，但同时也带来了许多工作流问题。 对于像 Atlassian 这样的厂商（Jira 和 Confluence），不能简单地将其收入按开发者席位（developer seat）来建模。实际上，Jira 和 Confluence 本身正在从单纯的人类开发者席位工具，转变为融合人类与 AI Agent 工作流的记录系统（system of record）。Jira 正在明显向管理人类与 AI Agent 的规划、治理和度量层（planning, governance, and measurement layer）推进。 而像 Ansible 这类工具则有所不同。AI 产生了更多软件工程基础设施变更后，公司反而需要更多的自动化（automation）、配置管理（configuration management）、策略执行（policy enforcement）和可重复部署（repeatable deployment）。
来源背景：Strategy Executive，美国头部电信运营商（前任职） - 支持 AI 编码带来更多 review/测试/工单等下游工作，Atlassian 捕获工作流规模增长而非 seat 增长，支出价值不会下降
我看到的比较明显的趋势可以分为两类：一类是实现AI编码（AI coding）商业化变现的工具；另一类则是享受到AI带来的下游工作流拉动效应（downstream workflow pull-through）的工具。 第一类最直接的代表是GitHub Copilot、Anthropic的Claude Code，以及Cursor这类AI原生编码工具。它们直接拉高了单开发者支出，因此即使开发者人数不增长，AI许可证和用量本身仍能持续增长。 第二类包括Atlassian、GitLab、Harness以及Red Hat的Ansible这类工作流和DevOps平台。因为AI编码生成了更多的代码和PR，后续必然会带来更多的代码审查（review）、测试、CI/CD、部署、安全扫描、工单处理和自动化需求。因此它们捕获的并非单纯的席位增长，而是工作流规模（workflow volume）的增长。 举个例子，Atlassian就是一个典型代表。即使工程人员编制基本持平，如果每位开发者管理的工程项目、工单、AI生成的变更以及Agent工作流都在增加，那么公司对Jira和Confluence这类记录系统（system of record）及工作流平台的支出价值就不会下降。Atlassian自身也在将Jira拓展为面向人类与Agent的统一工作管理层。 如果总结我目前最明确的判断：AI原生编码工具无疑是第一波受益者；而工作流、CI/CD自动化和DevSecOps则是第二波拉动效应的受益者。真正面临压力的是那些纯按人类席位收费、功能单一、且无法对Agent、自动化或用量进行商业化变现的开发者单点工具（point tools）。因为未来开发者席位增长可能会非常缓慢，但机器生成的工作负载增长会非常迅速。
来源背景：Strategy Executive，美国头部电信运营商（前任职） - 支持 标题称 MCP 用户扩大 Atlassian 支出的速度快两倍
3. **Atlassian CEO Mike Cannon-Brookes on Why the SaaSpocalypse Is Wrong, MCP Users Expanding Atlassian Spend Twice as
来源背景：TMTB 目录标题 - 支持 CEO 认为 agent 把软件变成数据库的说法只适用于平庸功能任务，不适用协作重的企业工作流
Cannon-Brookes rejects host Harry Stebbings’ framing that agents turn software into mere databases, saying that holds only for “pedestrian” functional tasks (like buying a charging cable) but not for collaboration-heavy enterprise workflows. He says Atlassian operates one of the top three-to-five MCP servers globally, enabling agents to read and write the Teamwork Graph, but argues that if Atlassian’s interfaces stop evolving to solve customer problems, the company loses relevance to that experience. He envisions a mixed future of horizontal agents, vertical deep-task agents, and persistent interfaces coexisting — not one “Uber agent” or single magic button. As evidence interfaces still carry commercial value, he cites Confluence generating north of $1.5B ARR, growing over 25% annually at scale — large enough to be a standalone public company — despite writing text seeming commoditizable.
来源背景：Atlassian CEO Mike Cannon-Brookes（摘要转述）
打包和提价：ARPU 提升
判断：成立，而且是过去一年加速里最实在的部分；但它本质上是提价和交叉销售，不是更多人在用。
Teamwork Collection (Jira + Confluence + Loom + Rovo 打包，credits 是单品的 10 倍) 不到三个季度卖出 1,000 家、100 万 seats，管理层说客户升级的头号原因是 10 倍 AI credits。打包要求各产品 seat 数一致，只买了 Jira 的客户等于被带着补齐 Confluence 和 Loom。IR 说升级 TWC 的客户用户数多约 10%。"TWC 客户每实例付费 seats 是单品客户的 4-5 倍"是横截面对比，更可能说明大客户更爱买 TWC。
另外两条 ARPU 杠杆：Premium/Enterprise 版本销售同比约 +40%；云每年提价，Mizuho 测算最新一轮挂牌价 +7-10%。Jira 每人每月约 $7-8 (管理层两次说法不一)，CFO 认为还有很大提价空间。迁移本身也是提价：小客户持平到 +5%，大客户最高 2 倍，第一年大约 1:1，第二、三年逐步恢复全价。
需要注意的是 Service Collection：渠道商说它是唯一有起色的 collection，因为 JSM 不再单卖，续约被强制转入。
Thank you all for joining us today. As you've already read in our shareholder letter, we closed out Q2 with very strong enterprise sales execution and incredible momentum across our business. We surpassed $6 billion in annual run rate revenue, delivered our first ever $1 billion cloud revenue quarter, up 26% year- over-year and grew our RPO 44% year-over-year to $3.8 billion. We have strong momentum across our enterprise, AI and system of work transformations, and you can see this in our numbers. Customers are choosing us for their future in bigger ways and bigger numbers than ever before. Enterprises like Cisco, Expedia, Reddit and Synchrony Financial rely on Atlassian to power their most critical business processes and workflows. Rovo surpassed 5 million monthly active users of our AI capabilities. We're seeing firsthand every day how AI is transforming the way that work gets done, and we are directly benefiting as a business. When we look at the thousands of customers in our software teams using AI code generation tools, we found that they create 5% more tasks with Jira, have 5% higher monthly active users and expand their Jira seats 5% faster than those who don't use these AI coding tools. As I've said before, AI is the best thing to happen to Atlassian and the results we're seeing today are no accident. As a long-term focused company, we're now benefiting from years of thoughtful investment across product, R&D and GTM, which have positioned us to capture this moment. These investments are creating what we believe is a truly differentiated customer experience. First, the data and domain expertise living inside our Teamwork Graph, which is now well more than 100 billion objects and connections across first and third-party tools, enables Rovo to deliver real business value that's context aware and actionable for customers across their search, chat and agentic experiences. Second, our decade-long investments in enterprise-grade security, data governance, permissioning capabilities and compliance enable every organization to securely move work forward at scale while deploying these fantastic new AI capabilities with the trust that they need. We provide a system of work built on deep integration into customer workflows with that compliance, security and support that enterprises trust built in. Lastly, our unique distribution engine enables us to seamlessly deliver those incredible experiences to over 350,000 customers, including more than 80% of the Fortune 500 and 60% of the Forbes AI 50. Customers are realizing the value of our unified system of work and want to continue to partner more closely with us than ever before. And all of this is driving our results. We closed a record number of deals greater than $1 million ACV in Q2, nearly doubling year-over-year again as enterprises are choosing to standardize on the Atlassian system of work. In less than 3 quarters, more than 1,000 customers have upgraded to our main AI monetization driver, the Teamwork Collection, purchasing more than 1 million seats to get the best AI platform and many more AI credits for their agents. As I look across our business, two things are clear. We've never been more of a strategic partner to the biggest businesses in the world, empowering their AI and future of work transformations and our momentum is continuing to grow. This gives us confidence in our road ahead and our long- term opportunity. We are truly transforming how work gets done and solving the toughest human AI collaboration challenges for our customers, and we're doing it every day. We're pushing ahead with strong conviction, and I could not be more bullish about the massive opportunities in front of us as we advance our mission to unleash the potential of every team. With that, I'll pass over to the operator for Q&A.
Yes. So the teamwork collection, as you talked about, has only been in the market for about 3 quarters now. As we rolled out Rovo, our AI capabilities across our platform, you get a certain number of AI credits with your standard subscription. With teamwork collection, you get 10x the amount of AI credits relative to U.S. stand-alone Jira stand-alone confluence subscription. So it's very appealing for customers as they think about their kind of AI journey because when you talk to our customers, they don't want consumption base pricing, at least today for our products. I think it makes sense for certain other models. But for our products, they don't really want to pay on a consumption basis. So today, if you use AI credits above and beyond your allotment on your standard Jira subscription or Confluent subscription, then you start to pay us on a consumption basis. But again, if you upgrade the Teamwork collection, you get 10x the amount of A credit. So it's a way to kind of forgo that worry or monitoring of AI credits today. That's been the #1 driver of why customers want to upgrade or choose to upgrade the Teamwork Collection. A lot of success early on, it's our expectations. 1,000 customers have upgraded to Teamwork collections. It equates to 1 million seats sold. Those customers that have upgraded Teamwork collection have expanded their seats versus their stand- alone footprint on Jira by double or by more than 10 percentage points of seats expansion, which is great to see that drives additional kind of attach of Confluence in Loom and then ultimately, the 10x amount of AI credits. So there's a lot of compelling kind of package within Teamwork Collection.
And when you look at the Teamwork Collection, which is our primary AI monetization vehicle right now, again, north of 1,000 customers, 1 million seats inside the first 6 months of upgrades of existing customers, some new in there, but mainly upgrades. choosing to upgrade Y, you get roughly 10x the AI or Rovo credits when you upgrade. We have a lot of other economic advantages. And I would argue, MAU advantages for the customer, you get the same number of Confluence, Jira and Loom seats, which increases your collaboration broadly, but again, far more AI credits. TWC customers use twice as many who use -- consume twice as many Rovo credits per paid user as non-Teamwork collection customers, and they have twice as many active agents. Now they have more credits. So like, yes, well, they've got more credits, they've opted in.
Ittai, if I can just chime in at sort of a nonfinancial and high level. Look, we're very clear that the system of work is about a continued growth of Atlassian into the knowledge worker population, right? When we talk about unleashing the potential of every team, our mission for over 2 decades, I would stress the every team part. We've got the software collection, which does very, very well. The DX and a lot of other things encompass and a bit like a pipeline, like there's a lot of great areas of that software collection. We've also got the Service Collection. So we're seeing great growth in Service Collection across HR teams, Finance teams, other areas outside of its traditional market in IT and Operations teams, but that's certainly an area of growth for us. So there's HR and Finance teams, we're seeing a lot in the Service Collection, for example, and you've seen us ship a lot of features. We also launched Customer Service this quarter as an application within that to go after a new set of teams we haven't been able to get to as well, you would say. In the Core Teamwork Collection, look, as we said in our letter, we were seeing double-digit seat expansion, more than double-digit seat expansion compared to people who buy the stand-alone applications. And that's baked into the Teamwork Collection packaging, but it's also because of the AI offering and how it works, right, in terms of getting -- you get -- you equate your Loom Confluence and Jira seats along with your additions, which often gains expansion. But given the nature of our applications and our continued growth in business teams across an Enterprise sales, Marketing, HR, Finance, et cetera, what that licensing structure allows those teams to do is to collaborate more. And collaboration is a very sticky and kind of viral activity. So that's where you're seeing that expansion coming through in our MAU and in our AI MAU and also ultimately ending up in our NRR and RPO numbers in terms of long-term commitments from customers that, that is where they're seeing expansion across business teams of all struts.
Yes. I think that premise is -- or that latter premise of this crude bundle is probably a misconception. So Teamwork Collection for everyone that's a little less familiar, you do get Jira, Confluence, Loom. But really, when we think about it, it's the value of the whole Atlassian platform. I think that's probably the biggest thing when I think about the evolution of our company over the past several years, it's -- we've built a unified platform with a unified data layer, and we talked about the teamwork graph underlying that, and you get the whole value of the platform with Teamwork Collection. And when we talk to customers, they do call out the increased AI credits as the #1 driver, but they say, okay, like what's resonating with me is I understand the value of the whole Atlassian platform, all the automation capability, of course, the Teamwork Graph, which we talked about, the analytics underlying it and then the kind of a unified AI layer underneath all that. And especially when I add additional users, add additional teams, then I get compounding effects to this because, again, you're adding more enriching the Teamwork Graph over time as I add more users, more workflows, understanding their tools, their priorities. And that's, I think, what companies understand. And so when they adopt Teamwork Collection, you actually see them expand above and beyond even their initial Jira footprint. So you tend to see them consolidate from other point solutions -- they're adding about 10% more users when they do so. And then when you look at the Teamwork Collection kind of behavior -- Teamwork Collection customer behavior, those customers are driving 2x more Rovo credit consumption than those that are in stand-alone products and have 2x more active agents, not just built agents, but active agents versus those on stand-alone products.
Thanks, Ryan. No, the philosophy remains -- really remains the same here. And on the cloud revenue drivers here on the $49 million beat, maybe I'll start with the fact that and highlight the fact that the DC migrations to cloud were really in line with expectations and not a contributor to the $49 million beat. It's progressing well and in line with our expectations for FY '26 contributing approximately mid- to high single digits to the cloud growth revenue. So the 2 primary drivers here of the outperformance were really the cross-sell and seat expansion that we were seeing similar to what we saw in the cross sell side, we continue to see strong uptake in our collections across Service Collection and Teamwork Collection. Service Collection revenue growth accelerated in Q4, just a strong signal of the AI adoption that we're seeing and that Mike mentioned a little bit earlier here. In fact, we're seeing over 3x increase in Agentic automation in Service collection as customers are continuing to deploy more agents beyond IT teams in HR and legal and finance and marketing teams. So again, also great to see the recognition here being named a leader in Gartner's Magic Quadrant for conversational AI platforms for employee services. We also continue to see a lot of strong performance on the Teamwork Collection side. A good reminder that Teamwork collection is still the best vehicle for customers to adopt and unleash AI on the Atlassian platform. This is where customers are upgrading to increased Rovo credits, it's giving customers what they want in terms of the predictability as well as the ability to really load balance those tokens across their organizations, and we're seeing that drive our strong ARPU uplift here as well. So really, I think what this shows is that the customers are consolidating their collaboration stack onto the Atlassian platform, recognizing the value that the Teamwork Graph really brings here. And we do see that in the results with Teamwork customers having 4 to 5x more paid seats per instance than stand-alone Jira or Confluence customers.
Sure, mate. I can take that one. Joe might want to add on at the end there. Look, the decision to embed Rovo in the core products, I would say, look, we are seeing continued growth and adoption of Rovo on our AI capabilities, right? As you called out, past 1.5 million monthly active users, continues to grow really strongly. Sales of premium enterprise editions are up over 40% year-on-year. And the cloud customers who have deployed Rovo agents to now tens of thousands of unique workflows connected to automation are seeing great value. So I would say that from that position of strength, we feel very good about the capabilities that we've built and we want to get them into as many hands as possible and as fast as we can amongst our user base, right? We believe that by maximizing usage and adoption, we -- yes, we've always been patient with monetization, but we think we have a massive user expansion opportunity, both in terms of additions and also growing wall-to-wall with more knowledge workers in those organizations. So that fits really well with the way that we think about building software in general. Again, we're always very thoughtful about balancing growth and capital usage. As we mentioned at the Investor Day, we've done some amazing work in R&D to make those capabilities able to be deployed at a reasonable cost, which is obviously a gateway in order to be able to make this sort of a move. The mindset you asked for were the decisions behind the decisions. Look, we do think that the future of teamwork is going to be about this sort of iterative human-AI, human-agent collaboration. We are, I think, very forward thinking in how we are doing that and implementing that. At the moment, we have a great offering. So we want to get that out to as many users as possible. We do see that it enhances our stickiness and promotes the wall-to-wall adoption of the Atlassian Cloud platform when this works. Again, we have a huge amount of organizational knowledge that we have access to, and giving them back to our customers in unique and interesting ways naturally makes them come back to our platform, which is a great advantage. And this really removes the friction to get more customers started on that journey, which is really important, that they can do that really quickly so that it can become a continued accelerant for Atlassian's continued growth in additions and more knowledge workers. And as I mentioned at the top, look, we have a really world-class R&D team that's built some amazing features in Rovo and all the capabilities, and we hence want to make sure that they come through to as many customers as possible. We have a pretty good history, I think, of making these sorts of bold business decisions, you might say, everything from PLG to free additions to moving to the cloud. So I feel really comfortable where we are coming from a position of strength and that we have the team that can follow through and execute against this to deliver the maximum amount of user value we can. Maybe I'll let Joe follow on, on the CAGR and the long-term nature from a financial perspective.
Mizuho says TEAM’s latest Cloud increases imply ~7–10% list-price uplift, modestly above last year, while expanded usage-based pricing could become a more material FY27 contributor. Mizuho estimates ~10% of Cloud customers may face incremental usage charges when billing begins Dec. 3, with penetration rising as Rovo/agentic AI adoption scales and potentially lifting ARPU through higher-tier Teamwork Collection adoption. Mizuho reiterates Outperform and raises PT to $215 from $175, with TEAM remaining a top pick.
来源背景：Mizuho 卖方（TMTB 转述）
So I think a couple of things to keep in mind is, again, the data center cohort is a much smaller pool of customers, but with much larger seat counts behind them. So that's the first thing that I would point out. I talked about it being a multiyear journey, I had to move these customers over to the cloud. You get varying levels of price uplift to that you're pointing at, depending on how many users the customer has. So on the low end of town, it could just be a cost neutral or 5% uplift. On the high end of town, there could be upwards of 2x the price. So it really depends on kind of the number of users that a customer has. We've been hesitant to give kind of a direct ratio to apply because of that varying level of complexity. But I think that's something that we're thinking more about how do we help investors understand the level of uplift. I think the commonality between all these though, is that because these are the largest customers, we do have a more traditional enterprise negotiation process with them. We want to ease them over to the cloud. So in those instances where they have more significant uplift. We want to smooth that path out for them. So in year 1, it may be less prohibitive and, therefore, make it more cost neutral for them, so more of a 1:1 kind of ratio in year 1. But in years 2 and 3, you start to peel back that level of discounting. So you slowly ramp them to that full list price over time. So you kind of actualize that price benefit in years 2 and 3.
So I think a couple of things to keep in mind is, again, the data center cohort is a much smaller pool of customers, but with much larger seat counts behind them. So that's the first thing that I would point out. I talked about it being a multiyear journey, I had to move these customers over to the cloud. You get varying levels of price uplift to that you're pointing at, depending on how many users the customer has. So on the low end of town, it could just be a cost neutral or 5% uplift. On the high end of town, there could be upwards of 2x the price. So it really depends on kind of the number of users that a customer has. We've been hesitant to give kind of a direct ratio to apply because of that varying level of complexity. But I think that's something that we're thinking more about how do we help investors understand the level of uplift. I think the commonality between all these though, is that because these are the largest customers, we do have a more traditional enterprise negotiation process with them. We want to ease them over to the cloud. So in those instances where they have more significant uplift. We want to smooth that path out for them. So in year 1, it may be less prohibitive and, therefore, make it more cost neutral for them, so more of a 1:1 kind of ratio in year 1. But in years 2 and 3, you start to peel back that level of discounting. So you slowly ramp them to that full list price over time. So you kind of actualize that price benefit in years 2 and 3.
展开全部 64 条原文 (按时间)
- 支持 TWC 长期增长逻辑：超过 1000 万 Jira 用户的 attach 机会、版本升级、wall-to-wall 带来的新增 seats、AI credits
Yes. Great questions, Keith. So from Q4 guidance approach, overall, there's no change. And we've taken the same approach to our guide for Q4 that we've taken all year, which is we aren't accounting for potential impacts of macroeconomic uncertainty and execution risk related to our enterprise go-to-market sales motion. I'd highlight that in Data Center and Marketplace, there is greater variability just given revenue recognition and the underlying dynamics in that part of the business around Cloud migrations, multiyear deal mix and the transactional nature of Marketplace revenue. I'd also highlight that as you think about our FY '26 outlook, keep in mind, we will continue to take a conservative risk-adjusted approach to our guidance for FY '26, just as we have throughout FY '25. In terms of the Teamwork Collection, while the Teamwork Collection is not a driver to our Q4 guidance and it will have a limited impact on FY '26, we think it is a powerful part of our strategy in transforming Atlassian. And the reason we think it makes sense is because there's an untapped attach opportunity to over 10 million Jira users. And we think we have a real opportunity to drive long-term revenue growth with this through additional attach of Confluence and Loom to Jira, additional attach of premium and enterprise versions since all users must be on the same version, incremental new seats because we have an opportunity to drive wall-to-wall and displace alternative tools, and then lastly, AI credits will drive AI usage. And that's going to strategically strengthen our structural competitive position broadly, and that will provide an uplift to our overall business. So overall, we're very bullish on the opportunity for TWC, Teamwork Collection, to drive growth over the long term.
- 支持 云业务受益于提价和 premium 版本升级；premium/enterprise 版本销售同比 +40%
Yes. Thanks for the question. It was another great quarter on the gross margin front. As you can see, we posted gross margins of 86%. That was solidly better than what we guided coming into the quarter. In Cloud, the story remains the same, just more of it. We continue to benefit from price increases, upsell to the premium edition and, importantly, engineering-driven investments we're making to optimize Cloud infrastructure and support costs. And as I've said in the past, focusing on and managing Cloud COGS efficiently right now is particularly important given the expected growth in that part of the business and the opportunity and strategy we have with Rovo in the AI space. I'd also reiterate that this highlights one of the big advantages of the engineering investment model we have at Atlassian that we've talked about in the past, which is investing in world-class engineering talent that can work on unlocking efficiency improvements and cost to serve. And as I mentioned last quarter, some of the very best work in the company is happening in the space. So we believe there's more to come. So overall, we feel very good about the gross margin picture and how trends are shaping up. And we do believe those efficiency gains will be structural in nature, and we'll be able to maintain them going forward.
- 支持 Premium/enterprise 版本销售同比增长超过 40%
Joe might want to add on at the end there. Look, the decision to embed Rovo in the core products, I would say, look, we are seeing continued growth and adoption of Rovo on our AI capabilities, right? As you called out, past 1.5 million monthly active users, continues to grow really strongly. Sales of premium enterprise editions are up over 40% year-on-year. And the cloud customers who have deployed Rovo agents to now tens of thousands of unique workflows connected to automation are seeing great value. So I would say that from that position of strength, we feel very good about the capabilities that we've built and we want to get them into as many hands as possible and as fast as we can amongst our user base, right? We believe that by maximizing usage and adoption, we -- yes, we've always been patient with monetization, but we think we have a massive user expansion opportunity, both in terms of additions and also growing wall-to-wall with more knowledge workers in those organizations. So that fits really well with the way that we think about building software in general. Again, we're always very thoughtful about balancing growth and capital usage. As we mentioned at the Investor Day, we've done some amazing work in R&D to make those capabilities able to be deployed at a reasonable cost, which is obviously a gateway in order to be able to make this sort of a move. The mindset you asked for were the decisions behind the decisions.
- 支持 已有 70 种消费计价杠杆，Rovo 也带消费计量，AI 时代消费计价会更重要
Yes. This goes back to the point around what happens in the future? So what do you believe AI will help people do? Fundamentally, everything we've seen so far, all the evidence points to a place where more things get built, more content gets created, more software gets created, more people are able to create things than we were able to do before. Now Atlassian is in the business of teamwork. So how do you help people create these new things? How do you help people create new software? Is it going to be the same people as we used to serve before? No, and that's a good thing because it's expanding the number of people we can serve, the kind people we can serve, and we see that evidence already. So it shows up in the business today already. Jira Service Management is one of our fastest-growing business, over $600 million. And Jira Service Management gets used by a lot of different teams who have nothing to do with software development. These are HR teams, these are marketing teams. The teams that are actually super charged by AI, they use our products because it makes them better. And the licensing model, specifically, we have consumption-based pricing, and we've had this for quite a while. But I do think consumption-based pricing will get more interesting and important in an AI world because as people unlock more productivity and as people learn to do more things with AI, they will also consume more of these kinds of capabilities. So we already have 70 different kinds of consumption-based pricing levers built into the product in Atlassian into the product portfolio. And Rovo is, our AI offering, also comes with a bunch of consumption- based meters. And I think those things will get more interesting in terms of how does the business evolve. But fundamentally, as usage increases, the business is going to benefit from this overall.
- 支持 Rovo 的消费计价限额目前很宽松、更多是成本保护，以后会考虑变现，也会重新看 per-seat 价格
Yes, I would say a couple of different things. So our #1 priority with Rovo has always been to drive usage and adoption first and foremost. And so this decision is much more in that vein of trying to mainline it throughout that customer base, trying to drive that usage in adoption as much as possible. Thus far, we've had great success. The customer response has been really good. We have over 1.5 million monthly active users of our AI capabilities across the platform today. That's up 50% quarter-over-quarter from just the December period. Again, our thinking around this is to remove as much of the friction and barriers as possible for customers and users to embed those AI capabilities and their daily workflows and habits, right? As these habits become much more ingrained, our products become stickier, more relevant. It gives us much more flexibility over time to recoup some of that value back via different monetization levers. We've always had a very patient for revenue approach. This is very consistent with that approach. It's all about delivering value first and foremost to customers and users. And then again, there's different vectors that expose themselves over time. Part of the other reason you want to drive usage and adoption is also to learn as much as possible, right? This is a very fast-moving landscape. You want to learn as much about user habits as possible, identify which different capabilities are being used and then you iterate much quicker from there in terms of the feedback loops and how do you develop the different AI capabilities and potentially expose different monetization levers. I think an important piece of this was we also introduced elements of consumption-based pricing with Rovo and so there is certain limits that we've introduced. Today, they are very generous with the name of driving usage and adoption. You don't want kind of this friction or barrier to drive that usage in adoption. And then over time, do we continue to look at those limits, make them much more of a monetization focus. Today, they're much more about almost a cost protection measure. But I think over time, we'll continue to look at the consumption-based pricing. We'll continue to look at kind of the per seat cost and respond accordingly over there. But again, it's all in the name of delivering value for us, being patient and then recouping that value back from customers over time.
- 支持 推出 Teamwork Collection：把 Jira、Confluence、Loom 和 Rovo 打包
Sure. No, I think as you noted, as I spoke with kind of customers and partners myself, I think that was the most evident thing is I think customers recognizing kind of the enterprise focus as well as how the platform investments that we've made over the past decade have started to show up much more in a more pronounced way. Again, that's across the different portfolio, but also across the capabilities throughout our products. I would say, probably the biggest announcement that I think that people focused on is our decision to make Rovo, our advanced kind of AI capabilities embedded throughout the platform. And so therefore, they're going to be included in the premium and enterprise additions of our core products with the standard edition to come. That probably was the biggest headline that a lot of folks in the investment community came away with. The other big ones that Rob talked about were introduction of collections. And so what that means is more holistic solutions tailored for different personas. And so what we introduced was a teamwork collection, which is a core set of collaboration products geared for any knowledge worker within an enterprise. But think of that as Jira, Confluence, a product called Loom and Rovo capabilities. And then as Rob talked about, another big announcement was our introduction of Government Cloud and an Isolated Cloud offering, all in the vein of what I mentioned earlier of migrating customers over to cloud in the coming years.
- 中性 分析师转述：Team'25 上有客户预计几年内会涨价
Robert W. Baird & Co. Incorporated, Research Division And on the Rovo announcement, that was definitely the biggest surprise or buzz at the event. And subsequent to the event, I've asked just about every company I cover or talk to what they think about it and whether they would consider doing what you guys did and a lot of thinking about it. And so it may be that you guys are ahead of the game and what was it that drove you guys to just say, hey, we're going to give Rovo away for free? And I guess I use free with a lowercase F because ultimately, I have a feeling that customers are going to pay for it and some indeed at the event even said, yes, we're probably going to see a price increase in a couple of years. But if we're using it, we're getting value, but what was the calculus that went into that thought process?
- 支持 TWC 是同一个 SKU，Jira 和 Confluence 在各版本上授权资格和数量一致（只有 Jira 的客户等于要补齐 Confluence seats）
Yes. So there's -- basically, they have the same 1 SKU whereby they would have the same eligibility and quantities license effectively across all Jira and Confluence across all of the tiers. Yes.
- 支持 TWC 包含 Jira、Confluence、Loom；只有 Jira 的客户会加上 Confluence，预期 Loom 也会扩张
Yes. So firstly, when we look at like Teamwork collection, that is our strategy moving forward. So we will lead with like Teamwork collection, and we will lead with the collections in general. So that will be our de facto for our -- let's say, our high-touch customers, our enterprise customers, so that will be what we lead with for, again, our enterprise customers. So that will be our de facto motion. So for any net new motion that we have, we will be leading with those. For any customer that's currently in data center, that's what we will be looking to move them to, that's where -- again, where we're going to be prioritizing our developments. For example, for Teamwork collection, included in Teamwork collection is Jira, Confluence and Loom. That means that they currently have Jira. They're also then moving forward are going to have a Confluence and then there will be the opportunity for them to have the Loom, and we expect to see an expansion of Loom as they move forward in Teamwork collection. I would imagine that customers who are already in the cloud with us are going to now be -- and we've already seen this, we'll be looking at possibly moving from potentially a stand-alone Jira or Confluence and looking at moving to Teamwork collection with us as well. We launched in early April, and we have had some inquiries from customers already and expect to see an uptick in terms of inquiries and demand from customers as we move forward.
- 中性 Rovo 免费提供，不收费
Sure. So we made the bold decision that we were going to make Rovo available for customers, and we were not going to charge for that. We announced that in our team earlier in April. And as you rightly said, we now have 1.5 million monthly active users, a number that we track very, very closely and track the daily active usage as well. We are very maniacally focused in terms of ensuring that our customers are now going to start to adopt this solution. Within customer success, we have ring-fenced people, our CSMs who are going to be focused on having our customers adopt Rovo. In addition, we're working very closely with our partners who are similarly going to be working on use cases to have our customers adopt the solution as well. Atlassian plays the long game, wherein we get the solution into the organization. This is a strategy that the company has done many times over, and it has worked very well. Get the solution into the company and it, in general, spreads. And then fast forward 2, 3 and 5 years and the solution is pervasive across the organization. When we look to companies like HarperCollins, the productivity gains that they are receiving, a large automotive company that I can't name, in terms of the productivity gains that they are also receiving in terms of the number of hours on the production line that they are saving as well. It's considerable. And now as we look at, not only the productivity gains and the numbers of hours that we're giving back to employees, we're now also looking at how we, as an organization, can start to use AI within our sales organization more and more as well. Because obviously, it's critically important that we're going to be a poster child for this as well. But I -- it's still very early days for us in considering that we enhanced this at team, which is only 2 short months ago. So I expect to see considerable movements over the coming months ahead.
- 支持 客户整体切换到 TWC 带来的价格提升，部分由 AI 驱动
Again, I've mentioned a Series of Customer examples. We could keep going on with customer examples, to be honest. We had a very large gaming company, iconic gaming company choosing Teamwork Collection, again for AI and the things that come with it so. The Teamwork Collection uplift for customers moving to that across their seats is partially AI-driven for sure, right? We're seeing that. We're putting it at the core of the platform. So that's one really important vector to make sure to understand.
- 支持 Premium/enterprise 版本同比增长约 40%，被视作 AI 变现的一部分
Again, I mentioned that we had 5x token usage growth quarter-on-quarter. We have a 50%-ish now growth rate of AI tools. So that's all really heading in that direction of usage first, and we will -- we're very good at monetizing in the long term. I will say that premium and enterprise additions again, had a growth rate of about 40% year-on-year for those core offerings. So that's another reason where you're seeing AI monetization. So we are thinking about it. We're spending time on it.
- 支持 Rovo 的间接收益会体现在 seat expansion、版本升级和迁移上；FY26 指引中 Rovo 消费收入很少
Yes. Thanks, DJ for the question. I'll start, and then Mike, feel free to chime in on top. As Mike articulated in the shareholder letter and in his comments, we are very excited about the strategic opportunity AI and Rovo present for us. In terms of the revenue impact, you asked about, we've mentioned in the past that the big focus for Rovo from the very beginning was around deployment, usage and engagement, and that revenue and monetization were secondary, longer-term considerations. While we have some consumption-based revenue from Rovo in our FY '26 plan, our guidance assumes very nominal amounts just given the limits we set to incent adoption and usage of our AI services. From there, we do believe some of the broader benefits from Rovo to our overall apps and collection portfolio, such as seed expansion, upgrade to higher additions and migrations will play out over time. So -- and in addition to that, we are seeing, as Mike pointed out, very strong uptake on premium and enterprise addition already. And some of that is a function of the AI and Rovo value that we put in the product. So we're super pleased with the reception of Rovo's launch. It's been exciting to see the reaction from customers and the increasing usage and adoption. And our big focus continues to be around deployment, usage and engagement, and monetization later. And since the launch, all 3 of those things are tracking along really nicely. Mike, anything else you want to add on that one?
- 支持 TWC 把 Jira、Confluence、Loom 和 Rovo 打包，给 10 倍 AI credits
What we've done maybe a little bit different than everybody else is we've launched Teamwork Collection at our team events in -- earlier in April this year. So we took Jira, Confluence and Loom and consolidated that into one bundle called Teamwork Collection and then included Rovo in that and gave our customers 10x the Rovo or AI credits that they would have received if they just bought the product stand-alone. So from a go-to-market perspective in order to increase our AI consumption, we've given, one, our customers basically the right to use these products as opposed to focusing our sellers on purely going out selling stand-alone AI. And now we are really using our customer success organization to drive the consumption of these products, which is a very different model than what others in the market are doing. So when we're out there and you hear Anu talking about the 2.3 million monthly active users, that is really 2.3 million monthly active users, which will be a number that's going to ramp up. And then from a go-to-market perspective to monetizing these opportunities, our focus is on, number one, the cloud migration, which is clearly going to accelerate our AI usage. And secondly, Teamwork Collection, which is a huge component of everything related to AI for us. And then upgrade to our enterprise and our premium editions, which is also going to be fueling additional usage and consumption of our AI credit.
- 支持 TWC 上市不到 2 个季度，带来用户数两位数百分比增长、版本升级和竞品工具整合
And I'd argue those things become even more important as more software is created and more people have the ability to create amazing technology that changes our lives. AI is also directly driving demand for our cloud offerings. Customers are choosing to migrate to the cloud and they're upgrading to the Teamwork collection to take advantage of our AI-powered cloud platform. In fact, in less than 2 quarters since we launched Teamwork collection, we've seen it drive a double- digit percentage increase in users as well as upgrades to higher value additions and the consolidation of competitive tools as our customers standardize on Atlassian. We're putting world-class AI at the center of our platform and throughout our entire set of collections and apps. We've showcased our relentless pace of innovation just a few weeks back at our sold out Team '25 Europe event in Barcelona with AI still in the show. Of course, you can read more about all these announcements in our shareholder letter.
- 支持 Atlassian 在 Jira 中加入 AI coding agent 比创业公司做企业销售容易，CIO 愿向可信厂商付适度加价
Once the "AI revenue" line item proves to be margin-accretive, the multiple expansion will return. 4. AI-native startups eat incremental LTV with better prices Rebuttal: The "Bundle" crushes the "Point Solution." Startups have a distribution problem; Incumbents have a feature problem. It is infinitely easier for Atlassian to add "AI coding agents" to Jira than it is for a startup to build an enterprise sales force, get SOC2 compliance, and displace a trusted vendor. CIOs will simply pay a modest upsell to a trusted vendor (zero marginal CAC for the incumbent) rather than onboard a risky new vendor for a "cheaper" price.
来源背景：VIC 作者 jgalt（买方）及其引用的 X 帖 - 支持 不到 3 个季度，1000+ 客户升级 TWC，购买 100 万+ seats
Customers are realizing the value of our unified system of work and want to continue to partner more closely with us than ever before. And all of this is driving our results. We closed a record number of deals greater than $1 million ACV in Q2, nearly doubling year-over-year again as enterprises are choosing to standardize on the Atlassian system of work. In less than 3 quarters, more than 1,000 customers have upgraded to our main AI monetization driver, the Teamwork Collection, purchasing more than 1 million seats to get the best AI platform and many more AI credits for their agents. As I look across our business, two things are clear. We've never been more of a strategic partner to the biggest businesses in the world, empowering their AI and future of work transformations and our momentum is continuing to grow. This gives us confidence in our road ahead and our long- term opportunity.
- 口径不齐 TWC 客户相比单品买家 seat 扩张超过两位数，但部分是打包机制（Loom/Confluence/Jira 席位对齐）带来的
Ittai, if I can just chime in at sort of a nonfinancial and high level. Look, we're very clear that the system of work is about a continued growth of Atlassian into the knowledge worker population, right? When we talk about unleashing the potential of every team, our mission for over 2 decades, I would stress the every team part. We've got the software collection, which does very, very well. The DX and a lot of other things encompass and a bit like a pipeline, like there's a lot of great areas of that software collection. We've also got the Service Collection. So we're seeing great growth in Service Collection across HR teams, Finance teams, other areas outside of its traditional market in IT and Operations teams, but that's certainly an area of growth for us. So there's HR and Finance teams, we're seeing a lot in the Service Collection, for example, and you've seen us ship a lot of features. We also launched Customer Service this quarter as an application within that to go after a new set of teams we haven't been able to get to as well, you would say. In the Core Teamwork Collection, look, as we said in our letter, we were seeing double-digit seat expansion, more than double-digit seat expansion compared to people who buy the stand-alone applications. And that's baked into the Teamwork Collection packaging, but it's also because of the AI offering and how it works, right, in terms of getting -- you get -- you equate your Loom Confluence and Jira seats along with your additions, which often gains expansion. But given the nature of our applications and our continued growth in business teams across an Enterprise sales, Marketing, HR, Finance, et cetera, what that licensing structure allows those teams to do is to collaborate more. And collaboration is a very sticky and kind of viral activity. So that's where you're seeing that expansion coming through in our MAU and in our AI MAU and also ultimately ending up in our NRR and RPO numbers in terms of long-term commitments from customers that, that is where they're seeing expansion across business teams of all struts.
- 口径不齐 TWC 要求 Loom、Confluence、Jira 席位数和版本一致，这本身就会'带出'扩张
Ittai, if I can just chime in at sort of a nonfinancial and high level. Look, we're very clear that the system of work is about a continued growth of Atlassian into the knowledge worker population, right? When we talk about unleashing the potential of every team, our mission for over 2 decades, I would stress the every team part. We've got the software collection, which does very, very well. The DX and a lot of other things encompass and a bit like a pipeline, like there's a lot of great areas of that software collection. We've also got the Service Collection. So we're seeing great growth in Service Collection across HR teams, Finance teams, other areas outside of its traditional market in IT and Operations teams, but that's certainly an area of growth for us. So there's HR and Finance teams, we're seeing a lot in the Service Collection, for example, and you've seen us ship a lot of features. We also launched Customer Service this quarter as an application within that to go after a new set of teams we haven't been able to get to as well, you would say. In the Core Teamwork Collection, look, as we said in our letter, we were seeing double-digit seat expansion, more than double-digit seat expansion compared to people who buy the stand-alone applications. And that's baked into the Teamwork Collection packaging, but it's also because of the AI offering and how it works, right, in terms of getting -- you get -- you equate your Loom Confluence and Jira seats along with your additions, which often gains expansion. But given the nature of our applications and our continued growth in business teams across an Enterprise sales, Marketing, HR, Finance, et cetera, what that licensing structure allows those teams to do is to collaborate more. And collaboration is a very sticky and kind of viral activity. So that's where you're seeing that expansion coming through in our MAU and in our AI MAU and also ultimately ending up in our NRR and RPO numbers in terms of long-term commitments from customers that, that is where they're seeing expansion across business teams of all struts.
- 支持 CFO：价格仍明显低于同行，还有很大提价空间
Yes. Thanks, Raimo. From a pricing perspective, on the Cloud, we invest quite a bit in R&D, and we're consistently delivering a lot of innovation and value to our customers. And that fundamentally allows us the opportunity to increase prices over time commensurate with that value delivery. We may do that through pricing or packaging of premium SKUs or through list prices. In either case, our prices today remain significantly below many of our software peers and competitors across our portfolio. And because of that and the pace of innovation and value delivery on mission-critical workflows, we still feel we have plenty of headroom for further pricing. In terms of data center, we will ensure that any price changes on data center going forward fit into the deliberate and planful approach we're taking in providing the right incentives at the right time to help customers upgrade to the cloud. Overall, however, we believe we remain competitively priced just relative to the value we deliver and competitive alternatives in that space as well. So that's how we think about the pricing perspectives in terms of the interplay between Cloud and Data Center.
- 支持 接受 TWC 并用 Rovo 的客户扩张了 20%
Yes. And so as you said, we have 5 million monthly active users at the end of Q2. We took a very different approach than other companies. We chose not to sell Rovo. Instead, we said that any customer who is moving to the cloud with us will get Rovo, specifically any customer that moves to Teamwork Collection, which is what I call our hero motion. Teamwork Collection includes Jira Confluence, Rovo and Loom. They get 10x the Rovo credit. We have been very successful with Teamwork Collection. It is really propelling our -- overall, our Rovo revenue stream, let's say. We have over 1 million seats in the cloud, and we have 1,000 customers that have moved to Teamwork Collection since April of last year. That is beyond what we expected. And we are very happy with the traction that we have received there. Specifically, what we have experienced is customers who have moved to Teamwork Collection who have embraced Rovo, we have seen a 20% increase in addition -- expansion as they have embraced Teamwork Collection, and we have seen an uptick in the Rovo usage as well. Just in the month of February, we had a large semiconductor company here on the West Coast go live with over 40,000 seats. Following their go-live 10 days later, there was an expansion to the tune of 8,000, the 20%. And that trend, we continue to see. We are consistently seeing more and more use cases emerging from our customers around the world. One thing that we are doing to drive more usage and more monetization is every one of our Teamwork Collection transactions that we were doing has a use case for Rovo embedded into it from our sales team, which means that our customer success team at the time of closure of that transaction will then pick up the use case and work with the customer and our forward deployed engineers that are embedded with the customer to help them activate the use case so that they can then get the journey started. And usually, what we see is that this leads to the creation of 3-plus agents at each of our customers.
- 支持 TWC 已有 1,000 家客户，超出预期，云端超过 100 万 seats
Yes. And so as you said, we have 5 million monthly active users at the end of Q2. We took a very different approach than other companies. We chose not to sell Rovo. Instead, we said that any customer who is moving to the cloud with us will get Rovo, specifically any customer that moves to Teamwork Collection, which is what I call our hero motion. Teamwork Collection includes Jira Confluence, Rovo and Loom. They get 10x the Rovo credit. We have been very successful with Teamwork Collection. It is really propelling our -- overall, our Rovo revenue stream, let's say. We have over 1 million seats in the cloud, and we have 1,000 customers that have moved to Teamwork Collection since April of last year. That is beyond what we expected. And we are very happy with the traction that we have received there. Specifically, what we have experienced is customers who have moved to Teamwork Collection who have embraced Rovo, we have seen a 20% increase in addition -- expansion as they have embraced Teamwork Collection, and we have seen an uptick in the Rovo usage as well. Just in the month of February, we had a large semiconductor company here on the West Coast go live with over 40,000 seats. Following their go-live 10 days later, there was an expansion to the tune of 8,000, the 20%. And that trend, we continue to see. We are consistently seeing more and more use cases emerging from our customers around the world. One thing that we are doing to drive more usage and more monetization is every one of our Teamwork Collection transactions that we were doing has a use case for Rovo embedded into it from our sales team, which means that our customer success team at the time of closure of that transaction will then pick up the use case and work with the customer and our forward deployed engineers that are embedded with the customer to help them activate the use case so that they can then get the journey started. And usually, what we see is that this leads to the creation of 3-plus agents at each of our customers.
- 支持 TWC 有 1,000 家客户、100 万 seats；首要升级动因是 10 倍 AI credits
Yes. So the teamwork collection, as you talked about, has only been in the market for about 3 quarters now. As we rolled out Rovo, our AI capabilities across our platform, you get a certain number of AI credits with your standard subscription. With teamwork collection, you get 10x the amount of AI credits relative to U.S. stand-alone Jira stand-alone confluence subscription. So it's very appealing for customers as they think about their kind of AI journey because when you talk to our customers, they don't want consumption base pricing, at least today for our products. I think it makes sense for certain other models. But for our products, they don't really want to pay on a consumption basis. So today, if you use AI credits above and beyond your allotment on your standard Jira subscription or Confluent subscription, then you start to pay us on a consumption basis. But again, if you upgrade the Teamwork collection, you get 10x the amount of A credit. So it's a way to kind of forgo that worry or monitoring of AI credits today. That's been the #1 driver of why customers want to upgrade or choose to upgrade the Teamwork Collection. A lot of success early on, it's our expectations. 1,000 customers have upgraded to Teamwork collections. It equates to 1 million seats sold. Those customers that have upgraded Teamwork collection have expanded their seats versus their stand- alone footprint on Jira by double or by more than 10 percentage points of seats expansion, which is great to see that drives additional kind of attach of Confluence in Loom and then ultimately, the 10x amount of AI credits. So there's a lot of compelling kind of package within Teamwork Collection.
- 支持 DC 迁移提价：小客户持平或 +5%，大客户最高 2 倍
So I think a couple of things to keep in mind is, again, the data center cohort is a much smaller pool of customers, but with much larger seat counts behind them. So that's the first thing that I would point out. I talked about it being a multiyear journey, I had to move these customers over to the cloud. You get varying levels of price uplift to that you're pointing at, depending on how many users the customer has. So on the low end of town, it could just be a cost neutral or 5% uplift. On the high end of town, there could be upwards of 2x the price. So it really depends on kind of the number of users that a customer has. We've been hesitant to give kind of a direct ratio to apply because of that varying level of complexity. But I think that's something that we're thinking more about how do we help investors understand the level of uplift. I think the commonality between all these though, is that because these are the largest customers, we do have a more traditional enterprise negotiation process with them. We want to ease them over to the cloud. So in those instances where they have more significant uplift. We want to smooth that path out for them. So in year 1, it may be less prohibitive and, therefore, make it more cost neutral for them, so more of a 1:1 kind of ratio in year 1. But in years 2 and 3, you start to peel back that level of discounting. So you slowly ramp them to that full list price over time. So you kind of actualize that price benefit in years 2 and 3.
- 中性 客户不想要消费计价，TWC 用 10 倍 credits 打消对 credit 用量的顾虑
Yes. So the teamwork collection, as you talked about, has only been in the market for about 3 quarters now. As we rolled out Rovo, our AI capabilities across our platform, you get a certain number of AI credits with your standard subscription. With teamwork collection, you get 10x the amount of AI credits relative to U.S. stand-alone Jira stand-alone confluence subscription. So it's very appealing for customers as they think about their kind of AI journey because when you talk to our customers, they don't want consumption base pricing, at least today for our products. I think it makes sense for certain other models. But for our products, they don't really want to pay on a consumption basis. So today, if you use AI credits above and beyond your allotment on your standard Jira subscription or Confluent subscription, then you start to pay us on a consumption basis. But again, if you upgrade the Teamwork collection, you get 10x the amount of A credit. So it's a way to kind of forgo that worry or monitoring of AI credits today. That's been the #1 driver of why customers want to upgrade or choose to upgrade the Teamwork Collection. A lot of success early on, it's our expectations. 1,000 customers have upgraded to Teamwork collections. It equates to 1 million seats sold. Those customers that have upgraded Teamwork collection have expanded their seats versus their stand- alone footprint on Jira by double or by more than 10 percentage points of seats expansion, which is great to see that drives additional kind of attach of Confluence in Loom and then ultimately, the 10x amount of AI credits. So there's a lot of compelling kind of package within Teamwork Collection.
- 口径不齐 Service Collection 是唯一有起色的 collection，原因是 JSM 不再单独售卖、续约被强制转入
SD: Yeah, so the collections, the service collection is really the only one that is gaining traction, and the main reason it's gaining traction is it's a forced buy. So if you're renewing your JSM, you have to renew it into the collection because JSM is no longer available on its own, it is only available from the service collection. So I don't know if they're going to do that pattern for the other collections. We haven't seen that yet, but I wouldn't be surprised if they did, if you could only get Jira in the other collection now, which name escapes me, maybe the software collection.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 支持 已有 Jira 生态降低 Jira Align 采用摩擦，体现生态交叉销售
EXPERT: We ultimately selected Jira Align because it provided the best balance of strategic portfolio visibility and agile execution alignment, which was the biggest gap we were trying to solve. While Planview was stronger for financial and capacity modeling, it was heavier, more expensive, and less intuitive for product and engineering teams who needed to use the system daily. Smartsheet, on the other hand, was useful for lightweight tracking but didn't provide the depth required for enterprise portfolio governance or end-to-end traceability. Jira Align fit best because it allowed us to connect strategy, portfolio, epics, delivery in a single system and improved cross-functional alignment across product and engineering. It also integrated well with our existing Jira ecosystem, which reduced adoption friction and time to value, even though we accepted some trade-offs in advanced financial planning capabilities.
来源背景：Mahak Varshney，Canadian Tire Product Manager - 支持 TWC 的 Rovo credits 是单品订阅的 10 倍，客户为此升级
Gregg, thanks for the question on that. Yes, on the cloud side, again, we saw strong performance reaccelerating to 29% year-on-year there. And maybe I'll start with the fact that DC migrations again to the cloud were in line with our expectations and not the realistic contributor to that $50 million beat on the print. It's progressing well, and we still expect that to contribute that mid- to single high-digit growth to cloud, like we mentioned. So the real 2 primary drivers that we talked about on the outperformance is really that cross-sell and seat expansion. . So on the cross-sell side, we saw outperformance in our collections business across Service Collection and in particular, Teamwork Collection. We have got Jira and Confluence, Loom and Rovo. And just keep in mind that right now, Teamwork Collection is really the best vehicle for our customers to buy and unlock AI and all the agent capabilities across the Atlassian platform, right? Customers are upgrading to TWC because of the increased AI credits. We're giving 10x more credits on Rovo versus the stand-alone subscription. And I'm sure Mike can touch a little bit more on some of the progress that we're seeing there. But importantly, we're also -- the other driver here is that we're seeing that growth in TWC while also seeing continued seat expansion in our core Jira stand-alone offering, right? And I think that really speaks to how an AI-driven agentic world, Jira and the Atlassian platform really remain core to enabling customers to manage their workflows and collaboration to fully unlock that value of AI. So whether it's TWC or standalone Jira offerings, we're launching a ton of new value that's really enabling our customers to deploy agents to do the work, to capture that agent activity alongside work history, full permissions and audit trails and admin governance. So a lot of great traction here, as you can see in our Q3 print.
- 支持 TWC 客户人均 Rovo credits 使用量是 2 倍以上，活跃 agent 数也是 2 倍以上
So we talk about that when people move to the Teamwork Collection, which we're seeing great momentum in. We talked about the cross-sell and the expansion earlier. The Teamwork collection gives you an amazing amount of software value across Jira, Confluence and Loom and all of the platform assets, goals and projects, et cetera, gives you access to Rovo, but it gives you an increased Rovo credit allowance. And we see that in the Teamwork Collection customers using more than twice as many Rovo credits per user, right? We want to make sure that we're building amazing features that use those Rovo credits, that the customers see value in using those credits. They also have more than twice as many active agents. So the Teamwork Collection comes with a larger pool credits, which customers are then using increasingly.
- 支持 客户从 DC 迁到云时仍有可观的价格提升
And maybe lastly, the cohort of DC customers that are actively planning and transitioning to the cloud, we're seeing these customers moderate their seat expansion. versus historical trends we've seen. Again, I'll mention that retention rate remains incredibly high, but now expecting a more muted level of data center expansion from these customers going forward as they try to move to the cloud. We're still seeing really nice uplift when customers move from DC to cloud. So net-net, what we're seeing is that our largest strategic customers continue to deepen their commitment at lasting, whether that's on DC or cloud, and we're working hard to meet them where they are and help them accelerate that transition so they can unlock all the AI and agent capabilities in the cloud. So hopefully that gives you a little bit of color there, Keith. And maybe I'll just share that with these dynamics sort of playing out across the year on data center, with Q4 yet to play out where revenue rec is being pulled into FY '26 from FY '27, we recognize that there's lumpiness in that pull-forward effect in data center and having the timing of -- and that does impact the timing of reported revenue RPO and CRPO.
- 支持 Teamwork Collection 成为主要 AI 变现载体（更大内嵌 credit 池、更多 agent 使用）
- Mgmt framed the upside as seat expansion plus cross-sell, not a one-off migration beat.- Rovo added millions of MAUs, AI Rovo credit usage grew >20% m/m, Rovo customers grew ARR ~2x non-Rovo customers, and Teamwork Collection is becoming the main AI monetization vehicle through larger embedded credit pools and more agent usage. The print weakens the “AI kills seats” thesis because Jira seat expansion and TWC upgrades both contributed to Cloud outperformance- Mgmt said it is seeing “the opposite” of seat compression, with strong expansion, strong cross-sell, strong AI usage, and NRR north of 120% with another sequential uptick.- Service Collection surpassed $1B ARR, grew >30% y/y, and mgmt called this TEAM’s largest-ever quarter for competitive displacements from a major ITSM provider (neg read-through to NOW)- Data Center Revenue $560.7M, +44.3% y/y vs Street $517.1M, +33.1%, benefited from ~$50M of greater-than-expected upfront term license revenue, helped by EOL dynamics, the large Q3 expiry base, and a March pricing change. Mgmt said Cloud migrations are on track and retention is robust, but Data Center customers preparing to migrate are moderating seat expansion, making FY27 DC deceleration steeper.- Non-GAAP OM was 34.0% vs Street 27.7%, and FY26 OM guide was raised to ~29.0% vs prior ~25.5% and Street ~25.6%. Bulls view this as evidence of Cloud platform leverage, restructuring savings, and expense discipline. Bears note the $50M upfront DC revenue benefit naturally lifts margin and Q4’s restructuring savings should not simply be extrapolated into FY27.- FCF $561M missed Street, with the miss tied to restructuring payments and billing/collection changes, including Data Center contract duration moving toward one year and annual billing on some multi-year Cloud contracts. Billings were also soft versus Street, but mgmt is trying to steer investors toward subscription ARR disclosures at the upcoming investor day to normalize revenue-recognition noise.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 支持 TWC 是主要 AI 货币化载体：前 6 个月超 1,000 客户、100 万 seats，主要是存量客户升级；升级后 Rovo credits 约 10 倍
I don't know if the 75% of the Fortune 500 companies that have turned on Rovo is a new stat, but it's there, and you can see our Agentic automations are growing really, really strongly. One way of using agents, again, is through automation, putting them into various rules that happen automatically and can access all of the Teamwork Graph. So some AI stats. And when you look at the Teamwork Collection, which is our primary AI monetization vehicle right now, again, north of 1,000 customers, 1 million seats inside the first 6 months of upgrades of existing customers, some new in there, but mainly upgrades. choosing to upgrade Y, you get roughly 10x the AI or Rovo credits when you upgrade. We have a lot of other economic advantages. And I would argue, MAU advantages for the customer, you get the same number of Confluence, Jira and Loom seats, which increases your collaboration broadly, but again, far more AI credits.
- 口径不齐 升级 TWC 后 Jira/Confluence/Loom seat 数不变——TWC 提升的是每 seat 价值/ARPU，而非 seat 数
And when you look at the Teamwork Collection, which is our primary AI monetization vehicle right now, again, north of 1,000 customers, 1 million seats inside the first 6 months of upgrades of existing customers, some new in there, but mainly upgrades. choosing to upgrade Y, you get roughly 10x the AI or Rovo credits when you upgrade. We have a lot of other economic advantages. And I would argue, MAU advantages for the customer, you get the same number of Confluence, Jira and Loom seats, which increases your collaboration broadly, but again, far more AI credits. TWC customers use twice as many who use -- consume twice as many Rovo credits per paid user as non-Teamwork collection customers, and they have twice as many active agents. Now they have more credits. So like, yes, well, they've got more credits, they've opted in.
- 支持 TWC 客户每付费用户 Rovo credits 消耗是非 TWC 的 2 倍，活跃 agent 2 倍（CEO 自承部分因 credits 更多）
choosing to upgrade Y, you get roughly 10x the AI or Rovo credits when you upgrade. We have a lot of other economic advantages. And I would argue, MAU advantages for the customer, you get the same number of Confluence, Jira and Loom seats, which increases your collaboration broadly, but again, far more AI credits. TWC customers use twice as many who use -- consume twice as many Rovo credits per paid user as non-Teamwork collection customers, and they have twice as many active agents. Now they have more credits. So like, yes, well, they've got more credits, they've opted in. Awesome.
- 口径不齐 AI 对 TWC 升级与云增长有贡献，但管理层承认难以拆分量化
Like it's a huge driver of that, right? It's a very obvious customer path to upgrade to Teamwork Collection to get far more AI Rovo credits. It's not the only reason to upgrade to the Teamwork Collection, but we're trying to make sure for customers that they see it as a simplified package. So there is some amount of that cloud growth that we are seeing and increased revenue from Teamwork Collection in the AI story. How much, it's hard to pass because there's a number of reasons where we've moved to Teamwork Collection, but it's always in the top handful of reasons. Why? Also, as we've talked about many times, the upgrades of the cloud, increasingly, AI is this is one of the top two reasons that customers will move, right? And so that -- the economics of customers moving to cloud for us is extremely good over that multiyear period. So both of those are actually leading to a positive economic outcome that's a result of our AI investments. We obviously have an AI Rovo credit story in the longer term. That requires that usage, that requires that flywheel to be moving. We're seeing it moving, which is great.
- 支持 CFO：TWC 是当下最好的 AI 货币化载体，credits 为单品 SKU 的 10 倍
And Mike touched on this a little bit earlier. But just maybe to put a finer point on it, TWC is the best AI monetization vehicle right now. And a big part of that is because of the credits that we're giving 10x credits relative to the stand-alone SKUs. But I think what we hear from customers is twofold. One is they want the access. Two, they also have within their enterprises, and I'm sure you guys see this within your organizations as well. There's lumpy usage. You got some hyper users, others that are provisioned but not using it. So they're able to load balance those credits across their organization as they're getting better and better at driving value from that. So it's giving them that access, it's giving them that predictability from an invoicing perspective. Then Mike shared a little bit earlier, all the different commercial models that we will adapt to as our customers are evolving as well. So I think we're seeing that. One of the things I don't want to get lost here is when we showed the video a little bit earlier there. I'm not sure if we ended up showing the stat, but in the keynote, we showed a stat that talked about how we were getting 44% of better quality in terms of output and 48% cheaper rate, right? You're using 48% fewer tokens to get to that same or better outcome. That's a conversation with your CIO. That's a conversation with your CFO that's going to resonate. And so I think both things are true. We're seeing that show up today in our monetization, and we're still building the platform in the flywheel to continue to compound on that.
- 中性 Rovo credits 目前未强制收费，管理层称走耐心路线，长期才有 credit 变现故事
Also, as we've talked about many times, the upgrades of the cloud, increasingly, AI is this is one of the top two reasons that customers will move, right? And so that -- the economics of customers moving to cloud for us is extremely good over that multiyear period. So both of those are actually leading to a positive economic outcome that's a result of our AI investments. We obviously have an AI Rovo credit story in the longer term. That requires that usage, that requires that flywheel to be moving. We're seeing it moving, which is great. We want to be very careful to grow, very patient, looking at that long-term story and making sure that it comes through. There is a story there for sure, but we're trying to make sure that, that is the case. And again, we'd rather have customers consuming and using and seeing the value of the platform. We will learn about the platform and really early in that journey as an industry, I think, everywhere.
- 支持 迁移带动更高价值 premium/enterprise SKU 采用
William Blair says Atlassian’s Team ’26 event reinforced the company’s long-term AI/platform opportunity, with GAAP operating profit growth expected to reaccelerate through 2027 as cloud migration, pricing, and infrastructure optimization improve margins. The firm highlights Atlassian’s “Teamwork Graph” and context engine as key AI differentiators enabling broader workflow automation and agentic capabilities across products. While Data Center revenue is expected to decline ~30% by 2027 as customers migrate to cloud, William Blair sees migration driving higher-value premium/enterprise SKU adoption and durable multi-year cloud expansion.
来源背景：William Blair（TMTB 转述） - 中性 客户可能抵触用量收费（为无意义任务或幻觉付 credit）
think about it now is if usage and seats stay, even better. It's even more sticky than before and there's more indicators in the platform of how strong it is in a customer's organization. There is a risk that customers could resist those consumption-based usage charges. If a customer feels they're being charged a usage credit for meaningless tasks or for hallucinations, for example, where the information is incorrect, this would be detrimental to the business. So time will tell. As I mentioned before in Monday's Q1 offering, though, 10% of its revenue in Q1 net new revenue did come from AI-enabled features. So this is a very interesting signal to watch.
来源背景：Jessica Saranich，前 monday.com Regional PMO Director & Chief of Staff - 支持 TWC 交叉销售超预期，ARPU 提升来自 TWC 货币化
Yes. Ultimately, I think the proof is in the performance, right? When I look at the outperformance that we had in Q3, it came from a couple of different areas. The first was cross-sell into our Teamwork Collection, and that outperformed our expectations. And that's really important because that's where -- it's the best vehicle to buy AI and unlock AI with Atlassian. So when customers think about AI right now, you can certainly buy single SKUs, Confluence and Jira and service -- our service management tools. But I think what we see over time in terms of the customer journey is that they go from single SKUs into a collection. And then they're seeing more and more value out of that collection and they buy more systems of work, right? So Teamwork Collection, Jira Service Management into Service Collection. And we unlock 10x credits for our customers. So it's the best vehicle for customers to actually unlock AI for their organizations right now. We're seeing that in our ARPU uplift when we monetize through Teamwork Collection and our other collections right now. So that's an area where we're seeing strong performance. We're seeing really strong traction, and that continues to play through. In terms of other areas of opportunity and maybe less so surprise, but we are still very early in the enterprise journey. We've been sort of evolving that go-to-market motion over the last several years, and it continues to mature. Brian and team are doing a fantastic job of that. And we shared some of these stats out at our investor forum a couple of weeks ago. The $1 million dollar customers cohort has grown 6x over the last 4 years, and they're growing at 39% year-over-year. And our $3-plus million cohort has 10xed over that same time period, growing at 54% year-over-year, and our fastest-growing cohort is our $10 million plus. And so as customers are continuing to adopt more and more AI and deploy more and more AI agents, we're continuing to take share in some of these really key markets. We talked about JSM, our Service Collection business, being $1 billion ARR business growing 30% year-over-year. Confluence is now a $1.5 billion business. Jira is a $2.5 billion business for us. And so a lot of opportunity to continue to grow into these different parts of the segments that we play in and all the different domains within those enterprises.
- 支持 TWC 已超 1,000 家客户、超 100 万 seats
Yes, it's hard to speak to sort of other companies and their decisions and how -- that they're making right now. But I think if I go back to the fundamentals of what's changing right now, right? We talked a little bit earlier about the barrier to entry on coding reducing, and that's going to continue to create more software, more innovation, more applications. And as a result, there's going to be a need for more planning, tracking and domain expertise, governance. And that's exactly where the Atlassian platform sits. And so when we saw the outperformance in Q3, as an example, that's customers telling us, listen, we want to unlock more AI with Atlassian. That was the outperformance we saw in Teamwork Collection. And we've got now over 1,000 customers in Teamwork Collection with over 1 million seats as part of that, and that continues to grow. And the other thing that we're seeing then is that it's allowing more domains to play in this space of creating more opportunity, more software, more innovation. The other piece I'm really excited about then is the Teamwork Graph, right? And we showed this at our Team '26, where we put Claude Code and we demonstrated 2 versions, one with Teamwork Graph and one without. Gave it the same repo, gave it the same prompt. And ultimately, we're able to show that with Teamwork Graph, we're able to demonstrate 44% higher quality outcomes at 48% fewer tokens. So that's an opportunity for us to continue to deliver value to our customers, right? Not all AI is created equal. We're demonstrating that through Teamwork Graph right now. And we have the ability, because we're also an open platform agnostic to models, to be able to optimize those models as well. There's a lot in that zeitgeist right now about token maxing and talking about making sure companies are responsibly managing those token costs. And I think Atlassian can play a big part of that.
- 支持 TWC 客户 credits 使用 2x、agent 2x；Rovo credit 使用月增 20%
I think different companies are on different arcs of that right now. I don't think it's one size fits all. I think what we're hearing writ large from our customers is, as it relates to AI, is they do want to unlock AI. We're seeing that in our performance in things like Teamwork Collection, where they get 10x more Rovo credits and the growth and opportunity that we see there. But they're looking also for predictability, right? We're hearing a lot right now about token maxing and making sure folks are being really mindful about the tokens that they're spending right now. And so they're looking for predictability. And they're also looking for the ability to load balance those AI credits as well, right? So for any enterprise and organization right now, oftentimes, what we see and what they see is that they'll see hyper users of those credits, and then those that are deployed, but not necessarily using those credits. So it's helpful for them to be able to take a package of credits and tokens and being able to load balance that across their organization as well. Right now, for Atlassian, the best vehicle to monetize AI and the best way for our customers to unlock AI is through Teamwork Collection because when they buy into that collection, they are getting 10x the Rovo credits. And we're seeing a lot of traction with that right now. We see that Teamwork Collection customers are using 2x more credits than those on a stand-alone basis. They're deploying 2x more agents, which then in turn consume more credits as well. And for Rovo customers, we're seeing credit usage grow 20% month-over-month. So a lot of traction on the AI side there. It's being monetized, and we're seeing that today through ARPU in our Teamwork Collection. And I think over time, that may shift over more to consumption, along with seats. And we also introduced our Flex program as well. This is a single wallet for customers, our largest really strategic customers to go all in on the Atlassian platform. That's access to the entire suite across our collections, across all of our applications and being able to unlock all of those credits across all of the different applications and load balance against those applications in terms of the value. And so through Flex, they're able to actually drive the type of value that is most meaningful for them across the entire platform.
- 反对 用户数下降时无法靠提价 50% 维持 per-user 收入，客户在谈判中占上风
**Expert:** I honestly cannot think of a world where it wouldn't work. I think that you can prove it with… I think the consumption-based model is the only one that's going to work for any company that's in SaaS at this point, in my opinion, because naturally, you're going to have less developers doing things. You need less people to do things now with the tools at hand. Most of the budget right now is going towards the large LLM models, Anthropic and OpenAI. Really, when you move all that money of a company to those purchases, then you have less money to spend on seats on GitLab or any other tools. Company is naturally going to figure out ways to cut down the number of users or if you could pivot to consumption model, you can actually keep those contracts steady and you don't have to churn in the next year or two at renewal. The consumption model is basically a way to financially… Accounting-wise, you are financially redoing the contract so that the customer doesn't feel like they're not getting the value add. Because if they lose users and you're still on user pricing, you can't convince them to pay 50% more per user. That's just not going to work. The cost of switching now is so much easier than before. The customer is always going to win that negotiation. If you move to consumption, it's like you can find a different way of modelling the next contract at renewal of like, "Hey, yes, you have less users, but here's how much more you're using. This is why this new pricing model works for you." It's easier to justify to a CFO who's trying to sign the contract. **LF:** Would you expect GitLab to price by tokens, tasks or some other measure within the new pricing model? **Expert:** It's going to be tokens. I think they haven't released what their strategy is, but it sounds like it's going to be tokens and very much like Claude and Cursor, all those other companies.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 反对 过去黏性高可在续约时涨价 50%，现在 AI 让迁移变容易，需要更多折扣
**Expert:** The tier down's biggest thing for GitLab specifically, it's the most expensive product is a security offering that goes on top of your normal standard platform. That security offering is no longer a value add anymore. That could cause some churn and tier down. Then, you have the web IDEs like Cursor, Claude Code doing all that work every day anyways. All you're doing is just prompting to create code into GitLab. You start questioning like, "Do we really need to pay for this many seats when you just have one developer doing all that by themself?" You can only pay for one developer seat of GitLab. I think those are the biggest risks. Then, you have other risks as just I'd say, maybe it's easier now to migrate off of GitLab or GitHub. It used to be very, very sticky. When I was there, any time we have a renewal, it was very easy to negotiate a renewal because you can literally just say, "Hey, you're already using our CI/CD, you can move off of GitLab, but it's going to take you two years to do it. It makes no sense." The cost of switching was so high, you can upcharge them 50% more on renewal and they wouldn't even want to move. They wouldn't even consider it to just take the 50% increase. Now, because of AI, it's actually pretty easy to migrate off of GitLab or everything is just easy to migrate in general. You really have to be prepared to win that customer and renew that customer without the threat of like, "Oh, it's just too sticky," because nothing is really sticky anymore. It's easy for anyone to just move off of anything these days.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 支持 客户案例：上云后从 3 个产品增至 9 个，支出提升 3-4 倍
Yes. So I think maybe the first thing to recognize is that enterprises have long operated in an ecosystem of multiple systems and multiple platforms. And Atlassian is one of those core platforms that sit within enterprises. So a couple of things really stand out when we talk to CIOs and C-suite customers here is that the System of Work, it is this journey. At the Investor Forum, we actually shared a couple of examples, whether it was a financial institution or an automotive company that's been an Atlassian customer for well over a decade, and as soon as they moved over to the cloud, they had roughly, call it, 3 products. This is 2022, 2023. When you fast forward 3 years later to where we are today, they're up to 9 products. They've tripled or quadrupled their spend with Atlassian. And what they're seeing is that as they moved from single SKU and moving into collections, right, again, the best way to adopt AI and unlock AI with those 10x Rovo credits is to actually purchase through Teamwork Collection. And so as they think about the platform and partners that they want to build long-term relationships with over time, Atlassian stands out as one of those. And another part of why that is, is because we are open by design, right? So it's not just the Atlassian platform that they're building on, but recognizing that large and strategic enterprises will build on top of multiple platforms. And the fact that we're open by nature, right, being able to connect to all the different platforms and applications that other -- that enterprises and customers have is a key differentiator for us. Because as the applications and workflows are being built on top of both Atlassian and others, it's writing back to the Teamwork Graph as well. So there's compounding value to customers, and they're seeing that as they go through their journey with Atlassian.
- 支持 TWC 带来的 ARPU 提升大于最初预期
Wells Fargo highlighted TEAM as one of its preferred 2H software ideas, citing improving seat growth, accelerating Teamwork Collection adoption, and growing AI-related developer activity. The firm noted 3Q strength was driven by Jira seat growth and Teamwork Collection upgrades, with Teamwork Collection becoming a larger ARPU uplift than initially expected. Wells also sees incremental upside from datacenter migrations, Rovo adoption, and stronger developer hiring trends, arguing investor concerns around AI have largely shifted from being a headwind to a potential growth driver.
来源背景：Wells Fargo 卖方（TMTB 转述） - 支持 客户升级 TWC 时整合其他点方案，用户数约增加 10%
Yes. I think that premise is -- or that latter premise of this crude bundle is probably a misconception. So Teamwork Collection for everyone that's a little less familiar, you do get Jira, Confluence, Loom. But really, when we think about it, it's the value of the whole Atlassian platform. I think that's probably the biggest thing when I think about the evolution of our company over the past several years, it's -- we've built a unified platform with a unified data layer, and we talked about the teamwork graph underlying that, and you get the whole value of the platform with Teamwork Collection. And when we talk to customers, they do call out the increased AI credits as the #1 driver, but they say, okay, like what's resonating with me is I understand the value of the whole Atlassian platform, all the automation capability, of course, the Teamwork Graph, which we talked about, the analytics underlying it and then the kind of a unified AI layer underneath all that. And especially when I add additional users, add additional teams, then I get compounding effects to this because, again, you're adding more enriching the Teamwork Graph over time as I add more users, more workflows, understanding their tools, their priorities. And that's, I think, what companies understand. And so when they adopt Teamwork Collection, you actually see them expand above and beyond even their initial Jira footprint. So you tend to see them consolidate from other point solutions -- they're adding about 10% more users when they do so. And then when you look at the Teamwork Collection kind of behavior -- Teamwork Collection customer behavior, those customers are driving 2x more Rovo credit consumption than those that are in stand-alone products and have 2x more active agents, not just built agents, but active agents versus those on stand-alone products.
- 口径不齐 Jira 约 $8/用户/月，低价高量；系统性提价
Yes. I think we take a lot of value in being a high-value provider at a more attractive price point. So that we think of it as almost high value, high volume. Core to our pricing philosophy has always been let's deliver value to the customer first and foremost. And then we can recuperate value through monetization over time. And so you've seen us at the price of Jira today is like $8 per user per month. It's incredibly affordable. But we've been very systematic with pricing. I think that speaks to the predictability aspect that we spoke about earlier. And again, we want to deliver that value. That's why we spend so much on R&D because we can deliver incremental value to customers every single year that affords us the ability to then take pricing over time because we are delivering more and more value to customers every single year. And again, I think that's core to our overall monetization model. And we want to maintain that kind of high-value, lower cost profile, especially because that's an enabler of being able to take share from customers or from other providers, I should say. Earlier, we talked about Teamwork Collection. When they adopt Teamwork Collection, we see a lot of consolidation off of other competing point solutions. That's going to be an important aspect, I think, in the years to come as you see more and more consolidation. And then especially when you think about the shift overall in the industry, I think that positions us really well as other companies also have to adjust to this landscape. Being a lower-cost provider, I think, will put us in a very advantageous position.
- 支持 Rovo AI 使用月增 20%+，Collections 仍是有吸引力的变现机会
Citizens maintained a Market Perform rating after channel checks, with 11 of 18 key data points coming in negative, led by evidence that Jira is losing share to Linear while the wind-down of Atlassian’s Data Center business remains the largest structural headwind. The firm highlighted encouraging momentum in Rovo, where AI usage is growing 20%+ m/m, and believes Collections remains an attractive monetization opportunity, but FY27/FY28 EPS estimates remain below Street. Citizens argues the stock’s 2.6x CY27 EV/revenue valuation appropriately reflects slower growth and competitive pressures.
来源背景：Citizens 卖方（TMTB 转述） - 支持 TWC 客户相对原单品 seat 扩张 >10%；专家认为打包红利会持续较长时间，但无法区分可持续部分
Teamwork Collection customers are reportedly expanding seats by more than 10% relative to their prior standalone product footprint.……It's hard to say because the initial bundling uplift is going to last a while.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 打包让只用 Jira、Confluence 用竞品的客户更倾向一并购买 Confluence（竞品替代）
**BR** [00:46:24]: It's hard to say because the initial bundling uplift is going to last a while. You have, I mean, hundreds of thousands of customers that own Jira, a lot less than that, that own Confluence, so what you're getting is a bundle, which has a better price, better support model and even a better integration model than you had before. If you're a customer that was using a Confluence competitor and you were using Jira before… Jira and you use a Confluence competitor, they suddenly made it a lot more compelling for you to consider buying the Confluence as Confluence plus Jira rather than that competitor. I think that the Teamwork Collection still has long ways to go before customers run out of it. Then again, they are continuing to churn and add. I'd be watching the PLG in all honesty, if I'm looking at it from a financial metrics perspective, given that PLG is how they land customers, if you're looking for net new prospects coming in, that's where you need to look. Everything else is upsell. If you've got net new prospects coming in at a good rate, and you're upselling them, let's say, even 20-30% of those customers end up buying other products then over a five-year period, that's how I do the math.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 TWC 客户每实例 paid seats 是单品 Jira/Confluence 客户的 4-5 倍（可能有选择偏差）
Thanks, Ryan. No, the philosophy remains -- really remains the same here. And on the cloud revenue drivers here on the $49 million beat, maybe I'll start with the fact that and highlight the fact that the DC migrations to cloud were really in line with expectations and not a contributor to the $49 million beat. It's progressing well and in line with our expectations for FY '26 contributing approximately mid- to high single digits to the cloud growth revenue. So the 2 primary drivers here of the outperformance were really the cross-sell and seat expansion that we were seeing similar to what we saw in the cross sell side, we continue to see strong uptake in our collections across Service Collection and Teamwork Collection. Service Collection revenue growth accelerated in Q4, just a strong signal of the AI adoption that we're seeing and that Mike mentioned a little bit earlier here. In fact, we're seeing over 3x increase in Agentic automation in Service collection as customers are continuing to deploy more agents beyond IT teams in HR and legal and finance and marketing teams. So again, also great to see the recognition here being named a leader in Gartner's Magic Quadrant for conversational AI platforms for employee services. We also continue to see a lot of strong performance on the Teamwork Collection side. A good reminder that Teamwork collection is still the best vehicle for customers to adopt and unleash AI on the Atlassian platform. This is where customers are upgrading to increased Rovo credits, it's giving customers what they want in terms of the predictability as well as the ability to really load balance those tokens across their organizations, and we're seeing that drive our strong ARPU uplift here as well. So really, I think what this shows is that the customers are consolidating their collaboration stack onto the Atlassian platform, recognizing the value that the Teamwork Graph really brings here. And we do see that in the results with Teamwork customers having 4 to 5x more paid seats per instance than stand-alone Jira or Confluence customers.
- 支持 TWC 是客户采用 AI 的最佳载体，带来强劲的 ARPU 提升
Thanks, Ryan. No, the philosophy remains -- really remains the same here. And on the cloud revenue drivers here on the $49 million beat, maybe I'll start with the fact that and highlight the fact that the DC migrations to cloud were really in line with expectations and not a contributor to the $49 million beat. It's progressing well and in line with our expectations for FY '26 contributing approximately mid- to high single digits to the cloud growth revenue. So the 2 primary drivers here of the outperformance were really the cross-sell and seat expansion that we were seeing similar to what we saw in the cross sell side, we continue to see strong uptake in our collections across Service Collection and Teamwork Collection. Service Collection revenue growth accelerated in Q4, just a strong signal of the AI adoption that we're seeing and that Mike mentioned a little bit earlier here. In fact, we're seeing over 3x increase in Agentic automation in Service collection as customers are continuing to deploy more agents beyond IT teams in HR and legal and finance and marketing teams. So again, also great to see the recognition here being named a leader in Gartner's Magic Quadrant for conversational AI platforms for employee services. We also continue to see a lot of strong performance on the Teamwork Collection side. A good reminder that Teamwork collection is still the best vehicle for customers to adopt and unleash AI on the Atlassian platform. This is where customers are upgrading to increased Rovo credits, it's giving customers what they want in terms of the predictability as well as the ability to really load balance those tokens across their organizations, and we're seeing that drive our strong ARPU uplift here as well. So really, I think what this shows is that the customers are consolidating their collaboration stack onto the Atlassian platform, recognizing the value that the Teamwork Graph really brings here. And we do see that in the results with Teamwork customers having 4 to 5x more paid seats per instance than stand-alone Jira or Confluence customers.
- 支持 TWC 的 Rovo credits 约为 10 倍，推动升级和扩 seat，seat 增加推动 ARPU
Rob, look, as we said, we continue to see a great increase in Rovo assisted actions, which is probably a rough proxy for credit usage. It's not directly linear, but the 50% quarter-on-quarter growth gives you a sense of the customer adoption of Rovo across the areas. Secondly, as we've said that the Teamwork Collection with approximately 10x as many Rovo credits included is a great motivator inducement for upgrade in terms of hitting those limits and looking at upgrading the Teamwork Collection, both in terms of moving to the Teamwork Collection, but also expanding the number of seats, the reps of deployment. Again, as we talked about, as I see the Teamwork Graph working in most different types of teams, you see seat increases, right, which is, again, driving that ARPU number, if you're looking at it at a really macro level, that is happening there. We have a series of consumptive elements today. We continue to work on how we judiciously and prudently roll out consumption-based offerings. We talked about both the direct consumption-based offering for overages. And secondly, the Flex model, which we continue to work on to allow customers to opt in to the entire Atlassian platform with a sort of dollar-based commitment that they can use across various different collections and consumptive elements in a singular package, both of which are early in their cycle holds great promise for the future. But as always, we maintain our sort of long-term thinking position and working through there. Right now, the simplest thing I would say is that the Teamwork Collection and the Service Collection are the vehicles for AI monetization, right? Customers adopting those, driving the usage within those. That is as we've said, that's the vehicle to measure AI monetization and adoption and both are going very, very strongly.
- 支持 TWC 客户 AI credits 和 agent 约 2 倍，每实例付费 seats 是单品客户 4-5 倍
Rovo is used by more than 80% of the Fortune 500, assisted actions rose 50% q/q, and Rovo adopters continue to grow ARR at more than twice the rate of non-adopters. MCP Server and Teamwork Graph CLI users surpassed 1M after more than doubling q/q, while MCP calls increased more than 400%. * **Monetization** AI monetization is showing up first through higher platform usage, seat expansion, cross-sell and ARPU rather than a separately disclosed AI revenue line. Teamwork Collection customers use roughly twice as many AI credits and deploy twice as many agents, while carrying four to five times more paid seats per instance than standalone Jira or Confluence customers. Mgmt continues to view Teamwork and Service Collections as the primary near-term AI monetization vehicles. * **Context** TEAM’s core strategic pitch is that foundation models will commoditize, but enterprise context will not. Teamwork Graph now spans more than 200B objects and connections, with TEAM claiming up to 44% more accurate answers using 48% fewer tokens.
来源背景：TMTB 作者转述业绩会与 callback - 中性 AI 变现首先体现在 seat 扩张、交叉销售和 ARPU，而非单独 AI 收入线
* **AI** The quarter materially strengthened the “AI beneficiary” narrative. Rovo is used by more than 80% of the Fortune 500, assisted actions rose 50% q/q, and Rovo adopters continue to grow ARR at more than twice the rate of non-adopters. MCP Server and Teamwork Graph CLI users surpassed 1M after more than doubling q/q, while MCP calls increased more than 400%. * **Monetization** AI monetization is showing up first through higher platform usage, seat expansion, cross-sell and ARPU rather than a separately disclosed AI revenue line. Teamwork Collection customers use roughly twice as many AI credits and deploy twice as many agents, while carrying four to five times more paid seats per instance than standalone Jira or Confluence customers. Mgmt continues to view Teamwork and Service Collections as the primary near-term AI monetization vehicles. * **Context** TEAM’s core strategic pitch is that foundation models will commoditize, but enterprise context will not.
来源背景：TMTB 作者转述业绩会与 callback - 支持 TWC 让客户以更高价格获得 10x credits，带来更高 ARPU
And so with each base subscription, you get a certain allotted number of Rovo credits. And then as you approach those limits, then you upgrade to the Teamwork Collection where you get 10x the amount. And that's much more customer-friendly in this moment in time where I think customers want that predictability or value that predictability of. I get 10x the amount of credits across my entire user base at a higher price point, so we realize higher ARPU as a result instead of worrying about Jason is going to chew through x amount of credits and I have to worry about this power user. It's a fungible pool of Rovo credits spread across my entire organization at 10x the amount. So it's much more predictable and customer-friendly, and we're just meeting customers where they're at today. I think over time, consumption-based pricing or usage-based pricing becomes a bigger part of the story.
- 支持 TWC：每 seat 支付更高 ARPU 换取可预测性
Yes. I think that's where we're -- we try to meet customers where they're at, and that's why I think something like Teamwork Collection and that model of bundling 10x the amount of Rovo credits per user gives customers like that you're pointing out, a much more predictable model. You're paying a higher ARPU on a per seat basis, and then I get a much more predictable kind of load balance across my user base.
- 支持 专家认为 Atlassian 靠推动客户升级到高级版/企业版获益（未谈 seat）
专家：Atlassian采用订阅制，并提供高级版和企业版。其盈利方式是推动客户从基础版升级到功能更多的版本。
来源背景：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团；自称并非 Atlassian 专家 - 支持 专家转述：采用 Rovo 的客户 ARR 增速超过未采用客户 2 倍；Rovo 辅助操作量环比约 50%
专家：这是一个很好的问题。我无法给出完整答案，但了解其中一部分。Atlassian 的数据表现非常好，Rovo 辅助操作量环比增长约50%，增幅非常显著。此外，Atlassian 已进入《财富》500强企业中的80%，不过这一数据可能通过统计口径进行修饰，使其看起来优于实际情况。 采用 Rovo 的客户，其 ARR 增长速度超过未采用客户的2倍。这说明客户确实从中获得了收益，也会推动 Rovo 继续增长。目前我还不清楚它存在哪些局限，但这些局限应该很快就会显现。Atlassian 正以企业搜索和未来 AI 智能体为切入点，并将自身定位为工作系统，类似于 Microsoft Graph 作为 Microsoft Copilot 的基础，由 Atlassian 在后台完成所有繁重工作。
来源背景：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团；自称并非 Atlassian 专家 - 支持 Jira 约 $7/seat，价格低给了提价空间（原文 L170-171 有断句缺失）
And if you look in the market today, whether it's our standalone apps or in our collection, the amount of credits that we allocate out is fairly generous and that is to reduce the friction as it relates to adoption and usage across our customer. This really gets back to number one, keeping the customer first. Number two, driving value and ultimately making sure that we're being really thoughtful about the pricing advantage that we occupy today to be able to take share in this. If you look at Jared today, it's roughly $7 per got to be some of the best B2B enterprise value in the market today. So that gives us a lot of opportunity from a headroom perspective. If you put yourself in the customer's shoes, you have to think about the cost of seats and they're having to think about the cost of tokens and agent point. But over time, more and more, they going to have to think about that holistically. You think about the pricing advantage that Atlassian has bring together, the seats, I think value we going to bring today, we're already starting to see that and our customers are reacting and you're seeing that in our results.
- 口径不齐 分析师：DC 客户过去几年平均提价 15-20%，复利后幅度很大
And you know, same thing with the data center migration journey, right ? So again, the pricing increases have been somewhat idiosyncratic to you versus your ability to just generally take price. But I want to ask you, what do you feel is your scope and ability to take more price in the install base ? Because I think as a financial analyst, I say I see, hey, on average it's been in 15 to 20% price increases for the data center customer in the last couple of years. And those numbers can can can get big after compounding for five years. Right. So does that still leave you with optionality on price ? Yeah. Listen, I think when we go back to the core values of driving value for some of the WHO actually large part of the lift that we see across our our business is actually not necessarily driven by those list price that we're talking about here. It is customers opting to expand opportunity set across our customers, increasing their seat count. It's upgrading their additions from premium to enterprise.
- 中性 Rovo credits 使用增加会给毛利率带来压力
If you think about the investments required to stand up an enterprise grade cloud to be able to build a architecture and platform that can deliver teamwork graph today we're cresting some of that investment right now and we're able to reallocate some of those reason. We also have an incredible engineering team that's great at optimizing against a lot of the the platform optimism that we're seeing. And also one of the advantages is to be able to look at the seven model makers that we work with, the over 70 models that we use across the platforms, behalf of our customers and be able to optimize against that as well. So driving leverage across gross margin dollars as customers continue to use more and more and find value in our Rovo credits, we're going to continue to see some of that pressure show up on the gross margin side, but that's also why we focus more and more on GAAP operating margin profitability. So some of that flow through in Q4. You've seen us set a positive or two half GAAP operating margin for FY 27. And as we drive more value to our customers, having that accrue to the top line, being incredibly disciplined on the investment side of the world, I would actually see that more.
- 支持 最新 Cloud 提价约 7-10% 标价上涨，略高于去年
Mizuho says TEAM’s latest Cloud increases imply ~7–10% list-price uplift, modestly above last year, while expanded usage-based pricing could become a more material FY27 contributor. Mizuho estimates ~10% of Cloud customers may face incremental usage charges when billing begins Dec. 3, with penetration rising as Rovo/agentic AI adoption scales and potentially lifting ARPU through higher-tier Teamwork Collection adoption. Mizuho reiterates Outperform and raises PT to $215 from $175, with TEAM remaining a top pick.
来源背景：Mizuho 卖方（TMTB 转述）
是 secular 加速，还是迁移和提价
判断：管理层把 Q3、Q4 的超预期归给 seat 和交叉销售，这本身可信；但整体增速里有一块会持续好几年、最终会结束的迁移红利，而有机部分的指引并没有加速。
管理层的说法：迁移"符合预期"，不是超预期的来源，超预期来自 cross-sell 和 seat expansion。CFO 在 Citi 也说大部分增量不靠挂牌价，而靠加 seat、升版本、从单品升到 collection。NRR 120% 以上，连续几个季度往上走。
但同一批材料也显示：迁移每年给云收入贡献中到高个位数的增长 (FY25、FY26 都是)，DC 2029 年 3 月停服，管理层预计迁移在 FY28-29 加速；迁移后的大客户 3 年 ARR 涨约 1.75 倍，说明迁移红利会拖很长的尾巴。Q3 FY26 的 DC 收入多出约 $50M，是 3 月提价前客户提前采购。2025 年 10 月云指引上调完全来自迁移，剔除 DC 停售后有机增长指引反而下调 0.5 个点。Investor Day 上管理层撤掉了 3 年 20% 收入 CAGR 的目标，说 FY27 收入增速见底。
所以"seats + ARPU = secular 加速"在数字上还看不到：能看到的是 seat 稳定增长 + 迁移 + 打包提价的叠加，叠加里只有 seat 那一块是长期的。
Thanks, Ryan. No, the philosophy remains -- really remains the same here. And on the cloud revenue drivers here on the $49 million beat, maybe I'll start with the fact that and highlight the fact that the DC migrations to cloud were really in line with expectations and not a contributor to the $49 million beat. It's progressing well and in line with our expectations for FY '26 contributing approximately mid- to high single digits to the cloud growth revenue. So the 2 primary drivers here of the outperformance were really the cross-sell and seat expansion that we were seeing similar to what we saw in the cross sell side, we continue to see strong uptake in our collections across Service Collection and Teamwork Collection. Service Collection revenue growth accelerated in Q4, just a strong signal of the AI adoption that we're seeing and that Mike mentioned a little bit earlier here. In fact, we're seeing over 3x increase in Agentic automation in Service collection as customers are continuing to deploy more agents beyond IT teams in HR and legal and finance and marketing teams. So again, also great to see the recognition here being named a leader in Gartner's Magic Quadrant for conversational AI platforms for employee services. We also continue to see a lot of strong performance on the Teamwork Collection side. A good reminder that Teamwork collection is still the best vehicle for customers to adopt and unleash AI on the Atlassian platform. This is where customers are upgrading to increased Rovo credits, it's giving customers what they want in terms of the predictability as well as the ability to really load balance those tokens across their organizations, and we're seeing that drive our strong ARPU uplift here as well. So really, I think what this shows is that the customers are consolidating their collaboration stack onto the Atlassian platform, recognizing the value that the Teamwork Graph really brings here. And we do see that in the results with Teamwork customers having 4 to 5x more paid seats per instance than stand-alone Jira or Confluence customers.
Right. So does that still leave you with optionality on price ? Yeah. Listen, I think when we go back to the core values of driving value for some of the WHO actually large part of the lift that we see across our our business is actually not necessarily driven by those list price that we're talking about here. It is customers opting to expand opportunity set across our customers, increasing their seat count. It's upgrading their additions from premium to enterprise. It's actually going from single SKUs to collections and then multiple collections over time. So I think that again, gives us a lot of permission, both from a pricing perspective, but we also think this is a moment to take share at the more and more customers. Think about how to get value out of AI.
Yes. Thanks for the question. We saw very healthy cloud migrations in Q2. It contributed a mid-to-high single-digit impact to Cloud revenue growth rates. So given this, we continue to expect migrations to drive a mid-to-high single-digit contribution to cloud revenue growth for the full year. So happy with the progress, and it continues to go very well.
And then Arjun, I'd just ask -- I'd add one other point that while we expect more momentum on migrations in the short term, there is going to be variability in the pace of these migrations quarter-to- quarter, and they will take time to move, and we expect most customers will migrate as we get closer to the data center end-of-life date in March 2029. So we expect migrations to accelerate in the '28, '29 time frame.
So let's look at multiyears, let's take a 3-year period. Let's take 1,000-plus user customers. So only the large customers have moved, 1,000-plus users that have upgraded to the cloud, what happens in their 3 years after they move to the cloud? Well, they grow about 1.75x, 1.5 to 2x growth in the 3 years after they migrate. Why? All those things I talked about before, seat expansion, cross-sell, they can get focus. They can get talent. They can get DX, they can get the product collection. They can get the service collection with all the capabilities.
Yes, Keith, thanks for the question. So on the data center side, the Q3 revenue beat, as we mentioned, was primarily driven by recognizing greater-than-expected upfront term license revenue within that quarter. Since our announcement of the data center end of life back in September, we've had a couple of quarters now to really better understand some of the signals that we're seeing from our customers in terms of their buying behaviors, especially Q3 being the largest expiry base for us. So let me unpack that a little bit more for us here. So first, I would say that the migration to the cloud is on track and continues as expected. We're pleased with what we're seeing there. We still expect that to contribute mid- to high single digits on cloud growth. And second is that the retention rates on our data center business remain incredibly robust, in fact, actually outperformed expectations in the quarter. And third, for some of our largest customers with more complex migration, they remain committed to transitioning to the cloud. But it's going to be a multiyear journey for them, right? They've got a lot of deep customizations, change management. It's going to take these customers time, right? Often many of these have tens of thousands of users, some with over 100,000 users. So this category of customers, we saw a pull forward of purchasing and expansion activity into Q3 from future periods. And we also had a pricing change in March. That further catalyzed this dynamic. So as a result, that drove greater-than-expected upfront term license revenue recognized in the quarter. In fact, relative to our expectations for Q3, we recognized approximately $50 million more in upfront term license revenue. And that's some of the trends that we've been seeing since our announcement of end of life back in September, but more pronounced in Q3 given the size of the expiry base and the pricing catalysts that I mentioned. And maybe lastly, the cohort of DC customers that are actively planning and transitioning to the cloud, we're seeing these customers moderate their seat expansion. versus historical trends we've seen. Again, I'll mention that retention rate remains incredibly high, but now expecting a more muted level of data center expansion from these customers going forward as they try to move to the cloud. We're still seeing really nice uplift when customers move from DC to cloud. So net-net, what we're seeing is that our largest strategic customers continue to deepen their commitment at lasting, whether that's on DC or cloud, and we're working hard to meet them where they are and help them accelerate that transition so they can unlock all the AI and agent capabilities in the cloud. So hopefully that gives you a little bit of color there, Keith. And maybe I'll just share that with these dynamics sort of playing out across the year on data center, with Q4 yet to play out where revenue rec is being pulled into FY '26 from FY '27, we recognize that there's lumpiness in that pull-forward effect in data center and having the timing of -- and that does impact the timing of reported revenue RPO and CRPO. So internally, of course, we look at a variety of metrics to performance manage our business, including a really healthy ARR. So next week, we're going to be holding that investor forum at our Team '26 Conference and to help guide investors through the revenue recognition timing dynamics on the data center side. We'll look to enhance our disclosures and share historical subscription ARR, which will help normalize some of those timing effects and help everyone better understand the underlying strength of the overall business. So all up, we feel really good about our execution, the runway that we still have ahead of us. So more to share next week at the TEAM event and the investor forum.
So I tried to explain this, both overcoming a bulls***, and I'll give you some illustrative charts. I thought I'd say all that so you were listening, I see you were writing before I showed you the charts. So this is an illustrative revenue growth. Number one, we're expecting a trough in total revenue growth in FY '27. Number two, we expect that to significantly reaccelerate in FY '28 as we lap that data center effect going through. You can see why that 1-year acceleration is a credit to the engineering team, creates a challenge for the finance team. And given these dynamics, the 3-year 20% CAGR that we set 2 years ago is no longer a relevant target to anchor on. What is relevant and a much better measure of the strength of our business is ARR.
Number one, we're expecting a trough in total revenue growth in FY '27. Number two, we expect that to significantly reaccelerate in FY '28 as we lap that data center effect going through. You can see why that 1-year acceleration is a credit to the engineering team, creates a challenge for the finance team. And given these dynamics, the 3-year 20% CAGR that we set 2 years ago is no longer a relevant target to anchor on. What is relevant and a much better measure of the strength of our business is ARR. Again, this is illustrative, don't get out your rulers. So why is that? Well, ARR normalizes the effect of ASC 606, that revenue recognition and helps everyone, I think, much better understand the overall health of our business. So if you take our subscription business, so data center and cloud, put it together, do a lot of great mathematics that our finance team is able to do, you see a very different pattern. Here is our subscription ARR for the last 7 quarters. What you can see in the last 3 quarters is a pretty good pattern of acceleration of ARR at the overall level of the business.
展开全部 53 条原文 (按时间)
- 口径不齐 CFO 明确 TWC 不进 Q4 指引、对 FY26 影响有限：打包对增长的贡献是远期的，不是当期
Yes. Great questions, Keith. So from Q4 guidance approach, overall, there's no change. And we've taken the same approach to our guide for Q4 that we've taken all year, which is we aren't accounting for potential impacts of macroeconomic uncertainty and execution risk related to our enterprise go-to-market sales motion. I'd highlight that in Data Center and Marketplace, there is greater variability just given revenue recognition and the underlying dynamics in that part of the business around Cloud migrations, multiyear deal mix and the transactional nature of Marketplace revenue. I'd also highlight that as you think about our FY '26 outlook, keep in mind, we will continue to take a conservative risk-adjusted approach to our guidance for FY '26, just as we have throughout FY '25. In terms of the Teamwork Collection, while the Teamwork Collection is not a driver to our Q4 guidance and it will have a limited impact on FY '26, we think it is a powerful part of our strategy in transforming Atlassian. And the reason we think it makes sense is because there's an untapped attach opportunity to over 10 million Jira users. And we think we have a real opportunity to drive long-term revenue growth with this through additional attach of Confluence and Loom to Jira, additional attach of premium and enterprise versions since all users must be on the same version, incremental new seats because we have an opportunity to drive wall-to-wall and displace alternative tools, and then lastly, AI credits will drive AI usage. And that's going to strategically strengthen our structural competitive position broadly, and that will provide an uplift to our overall business. So overall, we're very bullish on the opportunity for TWC, Teamwork Collection, to drive growth over the long term.
- 反对 DC 增长主要靠提价，被上云迁移部分抵消：这部分增长属于定价驱动，不是 seats
Yes. Let me start with the Cloud question. So we definitely did see deals fault more back-end loaded in the quarter than we expected. That did have impact on the Cloud revenue growth rate that we delivered in the quarter. As you know, from a billings perspective, that still is healthy, and we expect that revenue to be recognized in Q4. We just recognized less in Q3 than we expected. From a Data Center perspective, I would just say growth in the quarter was primarily driven by pricing, and it was partially offset by strong migrations to cloud and, as you mentioned, fewer multiyear agreements. I would just highlight that other drivers for data centers such as customer retention, renewal rates and expansion were healthy. So overall, we felt it was a good quarter of fundamental underlying Data Center performance. I would say if you think about the impact of multiyear agreements, we do recognize revenue in the period that we signed those agreements based on the total contract value. So when the duration of those agreements is shorter on average, we do recognize less revenue in the quarter, and that was definitely an impact on the Data Center revenue performance that we had in Q3.
- 口径不齐 迁移对云收入增长的贡献指引是 3 年内中到高个位数：说明有一部分云增长是迁移带来的
Yes. I'd say, to answer your first question, we definitely do believe that we're going to see better migration contribution to Cloud revenue growth in FY '26 and FY '27, consistent with what our guidance was at Investor Day last year. We said that those contributions would be in the mid-to-high single-digit range over a three-year period and that we would see less contribution in FY '25 simply as a result of the roll-off of the server and the support. So that's how we're thinking about the migration piece. And the second part of the question, yes, the question of around Data Center in FY '26. It's really too early at this point to talk about FY '26 guidance. What I can tell you is from that perspective, we will -- we do expect to continue to have Data Center growth driven by pricing and expansion at the customers. That will be offset though, however, by increasing Data Center to Cloud migrations.
- 支持 主持人提出“seat 不变+消费计价=价量双顺风”，管理层回答 Absolutely
Yes. And I think what you're saying is we can make an argument, and I want to continue it slightly longer here, that the seat side might be fine. But in addition to that, you're adding an additional pricing lever that may actually give you further upside. So it's like a box, right? Like if we just have a line, like you're only growing so far, but now we're actually having area, right? Price and quantity could be going up because of this and that could actually be a double tailwind for the long- term growth of Atlassian.
- 中性 上云迁移的流失和扩张都好于内部模型
Sure. So Atlassian historically had a server business. Think of this as a perpetual license model. So we would basically have customers buy the server software and really only pay us a maintenance fee there on. So several years ago, we decided to switch to a cloud model for two reasons. One, technologically cloud would allow us to offer innovation that server technology would not; and second, we wanted to shift to a subscription model from the perpetual license model. And we began this journey in 2020 and we are in the -- I would describe it as the final leg of that journey. And we've been phenomenally successful with that transition. So we've done better than what we had internally modeled in terms of customer churn. We've done better in terms of what we expected expansion to look like. So the fundamental thing to understand about the business is that the shift from perpetual to subscription business is complete for us in terms of we have -- we no longer have a perpetual license model, so the end of life the server product and data center and cloud both have a subscription model. And customers who adopt our cloud products, like I said before, show higher degrees of expansion that is adding more seats as well as higher degrees of cross-sell, adding more products. And we have a lot more cloud products than we do server products naturally because server was the legacy business. And over the years, we've managed to successfully migrate a lot of our server customers. So we went from the business being entirely server in DC, well over 90% of our customers now around cloud and 90% of our customers now choose cloud. And the set of customers, I would describe our journey as the leg that is about to come is that we have a set of enterprise customers that are some of our largest and most complex customers that have regulatory rules, that have constraints that need them to run hybrid environments, those are the customers that we're focused on over the next couple of years to bring them over to the cloud platform. I would describe the migration overall as having been extremely successful and better than what we expected.
- 支持 seat 扩张是主要增长来源，另外还有版本升级、企业客户、交叉销售和定价几个杠杆
I think what's also interesting, when we look at our seat growth, our seat growth has been very consistent over the past 4 quarters. So we continue to see really healthy seat growth. What's been also really interesting is when you cut it by tech startups, right? You would imagine that tech start-ups tend to be the most tech-forward companies. Their seat expansion patterns have been consistent with the seat expansion patterns that you see with SMBs across other industries. So that's a really encouraging sign for me, which again, I think, shows you our ability to serve not just probably developers, but also other types of users. And then earlier, you mentioned our focus on enterprise, right? And so while we certainly drive a lot of our growth via seat expansion, we also have developed different growth vectors in terms of upgrading people to premium and enterprise additions upselling and focusing on these enterprises, cross-selling across the broader portfolio of products and then pricing, right, capturing it via consumption pricing or perhaps per seat pricing over time. So I think we feel really good about kind of the culmination of these different vectors and our ability to drive durable growth no matter how the world continues to evolve with Gen AI.
- 中性 DC 合同期限从 2 年缩短到 1 年，推动上云
So the change that Rob is mentioning is actually just another step in kind of a series of changes that we've made to further incentivize cloud sales and kind of disincentivized data center. And it's kind of a reflection of that confidence that we have to better serve these customers in the cloud. And so we really talked about government cloud, isolated cloud, this continual kind of march forward to be able to serve these customers in cloud. We've then taken steps to kind of actively make it a little more difficult for customers to kind of renew one data center. And so that's just another step forward, is limiting customers from previously 2-year contract length on data center to now just limiting them to one year in term. And so that played a small role within our Q3 on the data center business. And then there's kind of a trickle effect into the marketplace as a result, as customers weren't renewing on their marketplace applications for 2-year periods because they could only renew for one year.
- 支持 增长杠杆是多元的：paid seat expansion、cross-sell、upsell、定价、新客户；AI 进一步扩大机会
Yes. Thanks, Kash. This is Joe. Great question, and I'll start and then pass it to Mike for more detail. So the first point I'd make is we have not changed our approach from last year with respect to our guidance for Q1 and FY '26. As we discussed last quarter, our guidance approach for FY '26 is consistent with the approach we took in FY '25, which is we are taking what I believe to be a conservative and risk-adjusted approach to account for macroeconomic uncertainty and potential business disruption from the evolution of our enterprise go-to-market sales motion because we believe those risks are still very relevant to the current operating environment. In addition, I'd say there's no change to our outlook that we will be able to deliver 20% compounded annual growth from FY '24 through FY '27. And we believe with 20% growth in FY '25, we have laid a strong foundation for achieving that and are on track to that goal. Our confidence in that remains rooted in the significant market opportunities we have in front of us. Our strategy to capture them through our investments in Enterprise AI and System of Work. And the strong momentum we've built around those strategies, as Mike laid out in the shareholder letter and in his opening comments. We have multiple levers of growth from paid seat expansion to cross-sell, upsell, pricing and new customer growth. And AI opens the aperture for even more opportunity given the structural position we have with our cloud platform and Teamwork Graph and deep investment in R&D. So taken together, the guidance approach and the opportunity for growth that we have, there's no change to the 3-year outlook we provided at our Investor Day last year, and we remain very optimistic about our ability to grow at a very healthy rate over a sustained period of time.
- 支持 Cloud NRR 120%
We reached 2.3 million AI now, up 50% from last quarter. And in less than a full quarter since its launch of Team '25, the Teamwork Collection has seen incredibly strong momentum and is exceeding our expectations. We are making significant progress on our key strategic priorities of serving the enterprise, delivering game hedging innovation in AI to our customers and connecting all teams through the Atlassian system of work. These big bets are paying off, resulting in a cloud net revenue retention rate of 120%. To further accelerate our progress, we are partnering with Google Cloud to bring our AI-powered Teamwork platform together with Google Cloud's AI optimized infrastructure. This partnership marks a major milestone in Atlassian's multi-cloud strategy, accelerating our cloud transformation and delivering advanced AI solutions to millions more users worldwide. In the AI-era, we believe the need for collaboration increases significantly as more people are able to create and more ideas can be brought to life.
- 口径不齐 FY25 迁移对云增长贡献中到高个位数（部分受 server 停止支持的推动）；FY26 指引贡献较低；未来 2-3 年中到高个位数
We believe there are 2 primary drivers behind these results. First are the investments we've been making to drive deployment and usage such as in our customer success team and in forward deployed engineering programs like Fast shift. And second, investments in R&D that continue to add differentiated value to our Cloud platform and apps through things like automation, analytics and AI, which makes the move to cloud more and more compelling over time. Altogether, that resulted in a contribution to Cloud revenue growth for FY '25 in the mid- to high single- digit range, which was also driven in part by the benefit of migrations related to server and to support in the prior year. Now going forward, we continue to believe the remaining customers on data center will take time to migrate to cloud and so that's some of the conservatism in our estimate. And the reason for that is the complexity of these migrations. Our remaining Data Center customers are some of our largest customers with the most complex environments to migrate, very different from the one to 2-day migrations for our smaller customers.
- 口径不齐 变现重点依次是上云迁移、TWC、版本升级（偏一次性的迁移和打包杠杆）
What we've done maybe a little bit different than everybody else is we've launched Teamwork Collection at our team events in -- earlier in April this year. So we took Jira, Confluence and Loom and consolidated that into one bundle called Teamwork Collection and then included Rovo in that and gave our customers 10x the Rovo or AI credits that they would have received if they just bought the product stand-alone. So from a go-to-market perspective in order to increase our AI consumption, we've given, one, our customers basically the right to use these products as opposed to focusing our sellers on purely going out selling stand-alone AI. And now we are really using our customer success organization to drive the consumption of these products, which is a very different model than what others in the market are doing. So when we're out there and you hear Anu talking about the 2.3 million monthly active users, that is really 2.3 million monthly active users, which will be a number that's going to ramp up. And then from a go-to-market perspective to monetizing these opportunities, our focus is on, number one, the cloud migration, which is clearly going to accelerate our AI usage. And secondly, Teamwork Collection, which is a huge component of everything related to AI for us. And then upgrade to our enterprise and our premium editions, which is also going to be fueling additional usage and consumption of our AI credit.
- 中性 刚宣布 Data Center 停止服务（EOL）
Yes. So firstly, we are increasing our spend this year in sales and marketing. You will see that over the course of the year, which means that we are literally hiring more people, both from a sales perspective and then from a customer success perspective as well, which is great news for us. We have invested heavily in terms of our own internal systems as well. We have revitalized our own leadership across the board, specifically in terms of the ecosystem, where I would say we had a little bit of a distance between ourselves and the ecosystem. Now we clearly have aligned goals between what's important to us and what's important to all of our partners, which is incredibly important because they do a significant portion of our revenue for us as well. Then at the same time, we have clearly incentivized our field from a sales perspective in the right way. I would say we had some room for improvement there from when I came in. We recently made an announcement around the end of life for data center just -- it feels like a couple of days ago, it was only yesterday. And that's obviously a big change for us. We were ready for this one. We have invested heavily around business value advisory, whereby we're going to be partnering with our customers now to really truly unlock the value of moving to the cloud and what this means for our customers. This is something from my own past with what I did around SAP and RISE. This is something that we, I'd say, learned a lot from and have some of the scars, but this is something that we are, I would say, more ready for at Atlassian than in my prior life. And we have invested ahead of the curve, so we're ready to go. And in addition, we have worked very closely with our engineering teams so that we set up a team called FastShift, which helps our customers move quicker to the cloud, which also helps train the ecosystem as well. That, in conjunction with training our sales teams has set us up for -- come out of the gate quicker in FY '26. And then maybe the last thing that I would just say is that we're really focusing on execution. Execution is the difference between good and really being awesome. And that's something that I've been pleasantly surprised with how well we have done in Q4 and expecting us to do a lot of good things in FY '26.
- 反对 云收入指引上调 1.5pt 只反映迁移和 Q1 超预期，其他有机驱动没上调
Yes. Great question. Thanks. And I'll try and clarify. So we are raising our cloud revenue outlook by 1.5 points to 22.5% year-over-year. That is only to reflect the stronger migrations performance and the outperformance in Q1. So we now expect migrations to make a mid- to high single-digit contribution to cloud revenue growth in FY '26. And for that migration upside to land in the back half of the year, just given the data center expiration base. And to your question directly, it's important to note we haven't made any changes to our other organic drivers of cloud revenue growth in our guidance. So we continue to maintain a conservative and risk-adjusted approach on all those other variables in the cloud and from a cloud revenue growth driver perspective for the rest of the year.
- 反对 剔除 DC EOL 后，有机收入增长指引下调 0.5pt（迁移导致确认节奏变化和 Marketplace 抽成降低）
Keith, this is Joe. I'll go first, and Mike will follow on with additional context. I'd say the key development that we saw in Q1 was that we had significantly stronger-than-expected cloud migrations from data center, which is a great thing for our business. As you know, this has been a big area of investment for us. It's a key strategic priority. It allows us the opportunity to provide more value to customers in ways that we simply can't on data center, and we can offer capabilities such as automation, analytics and AI, and so basically, it's definitely the most valuable and secure experience we can offer to our customers. So it's a great thing for the business. Cloud migrations also, however, have an impact on the timing of revenue recognition because cloud revenue is recognized ratably and data center has a combination of upfront recognition and some ratable. And then lastly, the move to cloud also impacts marketplace revenue because we have a lower take rate on cloud app sales than we do on data center apps. All other organic growth drivers in our business in Q1 were either slightly better or in line with our expectations. And we've maintained that same guidance approach on those factors that we held 3 months ago. So we continue to hold a very conservative and risk-adjusted outlook for all the other variables in the growth equation outside of migrations. So with that Q1 performance and momentum, we've adjusted our full year outlook for a greater cloud migration forecast. And we've left all other organic driver assumptions in place. And because of those revenue recognition timing differences between cloud and data center and the impact to Marketplace, this drives the 0.5 point decline in our organic revenue growth outlook for the rest of the year. And so that's the math underneath the guidance that we're giving for the full year on the organic part of the business. Mike?
- 口径不齐 DC EOL 定在 2029 年 3 月，预计迁移在 FY28-29 加速：迁移带来的是多年期但有终点的贡献
And then Arjun, I'd just ask -- I'd add one other point that while we expect more momentum on migrations in the short term, there is going to be variability in the pace of these migrations quarter-to- quarter, and they will take time to move, and we expect most customers will migrate as we get closer to the data center end-of-life date in March 2029. So we expect migrations to accelerate in the '28, '29 time frame.
- 支持 CFO：云增长驱动是更多用户、更高 ARPU、更多 cross-sell 和版本升级
Yes. Thanks, Karl. We do continue to have confidence in the long-term Cloud guide. If you think about the short-term guidance, we've taken the same approach in Q3 and FY '26 that we followed last year. The growth drivers for the Cloud business going forward continue to be very consistent with what we shared at the last Investor Day. We expect to tap into large market opportunities to drive healthy revenue growth through our strategies around enterprise AI and system of work. All of that drives a great number of users, a higher ARPU and more opportunity for cross-sell and upsell to higher value additions. And with AI, we believe we have a unique and differentiated position at this critical pivot point in the market with our Teamwork Graph around high-value mission-critical workloads, combined with our Cloud platform, and there's a lot of long-term opportunity in that space as well. So overall, those are the big drivers, and we continue to expect to drive healthy revenue growth over the next 2 years in Cloud.
- 支持 NRR 120%+ 且连续第三个季度上升
Thanks, Sanjit. Look, a totally valid and important question. Let me start with the numbers that you can see it, and then I'll talk to our philosophy. You can see in everything from the RPO to our NRR number, again, 120% ticking up for the third quarter and 120% plus, north of 120% and ticking up for the third quarter in a row. The pricing we currently have is delivering, right, for the customers. They are opting for more of what we are doing. And price is not a huge part of any of these conversations that we're having with our customers.
- 口径不齐 Q2 迁移对云增长贡献中到高个位数，全年预期相同
Yes. Thanks for the question. We saw very healthy cloud migrations in Q2. It contributed a mid-to-high single-digit impact to Cloud revenue growth rates. So given this, we continue to expect migrations to drive a mid-to-high single-digit contribution to cloud revenue growth for the full year. So happy with the progress, and it continues to go very well.
- 支持 预期多数客户会转向 Collection
Sure. Great question. So Atlassian is very interesting because like I said before, we entered into organizations and we entered into a department and then we spread within an organization. And now as customers have moved to Teamwork Collection, it's made the selling process for us within sales a lot easier because we've moved from product. And obviously, we're coming from a product-led growth motion to positioning outcomes. That's a journey that we, as an organization, are on to move more toward solution selling and less focus on the product motion. So as we position the Collection in total, we now focus on the outcome that we are driving for the business. We have champions within organizations that are within dev and IT. We now focus in on the business, and we talk to the business around the outcomes that we're going to be driving for them and how we can best support their particular area of the business. We have invested into our business value advisory team that helps us drive quantifiable results for the business in terms of dollar savings for the business as well. And at the same time, we also have a specific persona-based approaches that we also adopt, which is, again, new to Atlassian coming from a product-led motion. And when we look at the customers who have embrace Teamwork Collection, we have customers like Workday. Just recently in Q2, Workday, Tripadvisor, a large ride-hailing app, which I'm not allowed to tell you the name of them. And Publicis Sapient also and many, many more. As I said, 1,000 customers who have done this. And as we move forward, we expect the majority of our customers are actually going to embrace the Collection because that is the direction of travel, one of Atlassian, and that is where they received 10x the credits for Rovo, and that is where the majority of our customers are heading towards.
- 支持 上云后交叉销售和升级的机会更多
Sure. Great. So firstly, we obviously announced this, and we -- I think it was a matter of when we were going to announce this, not if we were going to announce this. There was not a massive amount of turmoil in the market when we announce it, firstly. We obviously have various cohorts of customers in terms of complexity who are ready to move to the cloud and then some that we need to work closely with, like I said at the beginning, because of their complexity. We have spent a lot of time with the ecosystem to ensure that they are ready to support us from a capacity perspective to move to the cloud, and I feel very good about where we are now and also good about the ecosystem and the hiring that they are making, which is why we have the incentives that we do in place because we will reach an inflection point here with more and more customers moving. We track very closely when we sign to when a customer actually starts that project to when it's actually adopted to the point that we're tracking monthly active users down to daily active users now to literally see where they are in their journey. And then when a customer does move to cloud, obviously, the experience that they have from everything from a user interface, it's completely different to innovation is completely different because the ability to consume and adopt Rovo is there in the cloud. Their ability to do so from an innovation perspective is really limited in data center. So they will experience something completely different. And then from a sales perspective, it opens up a lot more opportunities from cross-sell and upsell perspective for us. I will say, at the same time, there is a fair degree of work for us to be done for some of our larger customers, no doubt. However, we are also working with some customers in terms of isolated cloud already. And I have no doubt that we're going to continue to make progress in terms of supporting these more complex customers who are also leaning in heavily with us as well.
- 口径不齐 迁移提价第一年约 1:1，第 2-3 年逐步回到全价（多年期分摊、有限的一次性红利）
So I think a couple of things to keep in mind is, again, the data center cohort is a much smaller pool of customers, but with much larger seat counts behind them. So that's the first thing that I would point out. I talked about it being a multiyear journey, I had to move these customers over to the cloud. You get varying levels of price uplift to that you're pointing at, depending on how many users the customer has. So on the low end of town, it could just be a cost neutral or 5% uplift. On the high end of town, there could be upwards of 2x the price. So it really depends on kind of the number of users that a customer has. We've been hesitant to give kind of a direct ratio to apply because of that varying level of complexity. But I think that's something that we're thinking more about how do we help investors understand the level of uplift. I think the commonality between all these though, is that because these are the largest customers, we do have a more traditional enterprise negotiation process with them. We want to ease them over to the cloud. So in those instances where they have more significant uplift. We want to smooth that path out for them. So in year 1, it may be less prohibitive and, therefore, make it more cost neutral for them, so more of a 1:1 kind of ratio in year 1. But in years 2 and 3, you start to peel back that level of discounting. So you slowly ramp them to that full list price over time. So you kind of actualize that price benefit in years 2 and 3.
- 中性 DC 客户必须在 2029 年 3 月前迁走，是多年期的过程
Yes. So today, we have two deployment models. It's -- we have our SaaS offering or Cloud offering, and then we have what we call the data center offering, which is an enterprise-grade on-prem deployment model. We announced the end of life of that data center deployment. And so those customers will have to move off of that by March of '29, so a little over 3 years from now. And during that time, we'll continue to migrate those data center customers to the cloud. That's an important aspect for us because the cloud is where the innovation is. Earlier, I talked about the platform. That's where the lion's share of the investment has been. That's where you have AI capabilities, that's where you have analytics, that's where you have automation capabilities. None of that's available on data centers. So it's really important that we migrate those customers to give them the best experience that they can get, which is going to be on the cloud platform. These are our largest customers, the data center cohort, right? Like these are the biggest or big enterprises, many of them who have been Atlassian customers for over a decade. Many of them have 50,000 users or more. And so it requires a little more handholding to bring these data center customers over. They have customized kind of footprints on their data center instance. So part of the challenge and an opportunity, of course, is how do we get them to rebuild some of that customization onto the cloud platform. That's why we know it's going to be a multiyear journey for some of these customers because they have to rebuild some of the customization on the cloud instance. How do we hold their hand through it with more kind of white glove service? Like these are all the different avenues that we've looked at. We have a program now called Fast Track, which is giving them a dedicated R&D resource to help them road map out the path to get to the cloud. That has accelerated customers' migration plans quite considerably. And so that's something that we're rolling out with larger scale today.
- 支持 非 PubSec 超预期主要由 AI 加速编码驱动，其次是上云迁移管道超预期
SD: I think it is mainly the AI, the accelerated AI code that we're seeing more and more. It's like 6 months ago, most of
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 口径不齐 上云迁移是非 PubSec 增长的第二原因，管道超出预期（属一次性/阶段性来源）
SD: There has been a change from my normal stance on this, and we are seeing more. That's probably the secondary reason on our non-PubSec customers that we're finding more customers that want to go or that are kicking the tires of doing a data migration. So I didn't anticipate that we'd get a lot this year. There's already more in the pipeline than I anticipated.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 支持 迁移符合预期，不是 Q3 超预期来源
Gregg, thanks for the question on that. Yes, on the cloud side, again, we saw strong performance reaccelerating to 29% year-on-year there. And maybe I'll start with the fact that DC migrations again to the cloud were in line with our expectations and not the realistic contributor to that $50 million beat on the print. It's progressing well, and we still expect that to contribute that mid- to single high-digit growth to cloud, like we mentioned. So the real 2 primary drivers that we talked about on the outperformance is really that cross-sell and seat expansion. . So on the cross-sell side, we saw outperformance in our collections business across Service Collection and in particular, Teamwork Collection. We have got Jira and Confluence, Loom and Rovo. And just keep in mind that right now, Teamwork Collection is really the best vehicle for our customers to buy and unlock AI and all the agent capabilities across the Atlassian platform, right? Customers are upgrading to TWC because of the increased AI credits. We're giving 10x more credits on Rovo versus the stand-alone subscription. And I'm sure Mike can touch a little bit more on some of the progress that we're seeing there. But importantly, we're also -- the other driver here is that we're seeing that growth in TWC while also seeing continued seat expansion in our core Jira stand-alone offering, right? And I think that really speaks to how an AI-driven agentic world, Jira and the Atlassian platform really remain core to enabling customers to manage their workflows and collaboration to fully unlock that value of AI. So whether it's TWC or standalone Jira offerings, we're launching a ton of new value that's really enabling our customers to deploy agents to do the work, to capture that agent activity alongside work history, full permissions and audit trails and admin governance. So a lot of great traction here, as you can see in our Q3 print.
- 支持 使用 Rovo 的客户 ARR 增速约为未使用客户的 2 倍（可能有选择偏差）
Thank you all for joining us today. As you've already read in our shareholder letter, we delivered some incredible Q3 results. Total revenue grew 32% year-over-year to $1.8 billion. Cloud revenue surpassed $1.1 billion and accelerated to 29% growth year-over-year, and RPO grew again 37% year-over-year to $4 billion. These are likely thanks to our team's excellent execution and clear momentum across our key strategic priorities: enterprise, AI and the system of work. This quarter, some of the world's largest enterprises, including Siemens Energy, Rheinmetall and Wayfair deepened and broadened their commitments to Atlassian. In AI, we continue to add millions of monthly active users to Rovo and our AI Rovo credit usage is growing more than 20% month-over-month. Customers using Rovo are also growing their ARR at roughly 2x the rate of customers who are not using Rovo, contributing to our strong cloud outperformance and expansion in the quarter. And more and more enterprises are embracing our platform-wide vision using Atlassian system of work to see the full picture of their organization. This is because the Teamwork Graph connects knowledge, work, people and code, giving our customers one of the richest enterprise context graphs in the world. Context is a clear differentiator for us. And we're saying this is our competitive momentum built. This is our largest ever quarter for competitive displacements from a major ITSM provider. We're taking it from rivals as customers move away from legacy systems and choose Atlassian for a more modern AI native and much better value service platform. As I've said before, I believe AI is one of the best things that has ever happened to Atlassian. In a world where human will run teams of agents, context is the only anchor to avoid chaos. And we believe the companies that prioritizes context will come truly AI native. With Atlassian, our customers aren't just choosing software that choosing the kind of company that they want to become. This is a time of significant change in our industry, and we're moving forward with strong conviction and discipline. We're focused on executing, delivering customer value and driving, durable profitable growth. With that, I'll pass the call to the operator for Q&A.
- 反对 DC 收入超预期约 5000 万美元来自大客户提前采购和扩容，3 月提价进一步催化：一次性、跨期
They've got a lot of deep customizations, change management. It's going to take these customers time, right? Often many of these have tens of thousands of users, some with over 100,000 users. So this category of customers, we saw a pull forward of purchasing and expansion activity into Q3 from future periods. And we also had a pricing change in March. That further catalyzed this dynamic. So as a result, that drove greater-than-expected upfront term license revenue recognized in the quarter. In fact, relative to our expectations for Q3, we recognized approximately $50 million more in upfront term license revenue.
- 支持 NRR 保持 120% 以上，连续 3-4 个季度上升
Yes, Gregg, thanks for that and for calling it out. I hope we've been very consistent on our views in that world. We are not seeing any signal of seat compression from customers. If anything, we are seeing the opposite. We are seeing strong expansion numbers, strong cross-sell numbers between collections, strong usage of AI and strong commitment to the Atlassian platform. Many, many competitive wins, a huge amount of consolidation happening into the Teamwork Collection. So we have a lot of green lights in a lot of different places. Similarly, you can see that NRR maintained north of 120% and even ticked up again for, I think, the third or fourth quarter in a row. We have a lot of reasons why that is. Firstly, I would go to high R&D investment in just great quality software, right? That does matter. It always gets ignored in a lot of context, wrong choice of words, in a lot of quarters. But we build amazing applications that deliver great value for our customers. Secondly, the context we have in the Teamwork Graph and the critical business processes continue with AI to blur the boundaries between teams and between roles. That means our platform and our offering between service teams on one side that connect to finance and HR, marketing and in through ops teams to technology teams with business teams and leadership teams. The same context across all of those teams allows us to expand into those nontechnical roles at an increasing rate, right, which shows why we are getting that seat expansion in different collections, in different areas and just the continues in our business. Fundamentally, customers are opting for more and more workflows on the Atlassian platform.
- 支持 超预期被定性为 seat expansion+交叉销售，而非一次性迁移
- Mgmt framed the upside as seat expansion plus cross-sell, not a one-off migration beat.- Rovo added millions of MAUs, AI Rovo credit usage grew >20% m/m, Rovo customers grew ARR ~2x non-Rovo customers, and Teamwork Collection is becoming the main AI monetization vehicle through larger embedded credit pools and more agent usage. The print weakens the “AI kills seats” thesis because Jira seat expansion and TWC upgrades both contributed to Cloud outperformance- Mgmt said it is seeing “the opposite” of seat compression, with strong expansion, strong cross-sell, strong AI usage, and NRR north of 120% with another sequential uptick.- Service Collection surpassed $1B ARR, grew >30% y/y, and mgmt called this TEAM’s largest-ever quarter for competitive displacements from a major ITSM provider (neg read-through to NOW)- Data Center Revenue $560.7M, +44.3% y/y vs Street $517.1M, +33.1%, benefited from ~$50M of greater-than-expected upfront term license revenue, helped by EOL dynamics, the large Q3 expiry base, and a March pricing change. Mgmt said Cloud migrations are on track and retention is robust, but Data Center customers preparing to migrate are moderating seat expansion, making FY27 DC deceleration steeper.- Non-GAAP OM was 34.0% vs Street 27.7%, and FY26 OM guide was raised to ~29.0% vs prior ~25.5% and Street ~25.6%. Bulls view this as evidence of Cloud platform leverage, restructuring savings, and expense discipline. Bears note the $50M upfront DC revenue benefit naturally lifts margin and Q4’s restructuring savings should not simply be extrapolated into FY27.- FCF $561M missed Street, with the miss tied to restructuring payments and billing/collection changes, including Data Center contract duration moving toward one year and annual billing on some multi-year Cloud contracts. Billings were also soft versus Street, but mgmt is trying to steer investors toward subscription ARR disclosures at the upcoming investor day to normalize revenue-recognition noise.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 支持 1,000+ 用户的大客户迁云后 3 年 ARR 增长约 1.75x（1.5-2x），驱动包括 seat expansion 和交叉销售；管理层称迁移时并未一次性收取全部价值
So let's look at multiyears, let's take a 3-year period. Let's take 1,000-plus user customers. So only the large customers have moved, 1,000-plus users that have upgraded to the cloud, what happens in their 3 years after they move to the cloud? Well, they grow about 1.75x, 1.5 to 2x growth in the 3 years after they migrate. Why? All those things I talked about before, seat expansion, cross-sell, they can get focus. They can get talent. They can get DX, they can get the product collection. They can get the service collection with all the capabilities.
- 支持 迁移不是一次性项目而是飞轮，cloud NRR 持续 120%+
In fact, we're trying to trade that off over that 3-year period. And you've probably seen this chart before, but the cloud is not a flywheel, right? Sorry, it is a flywheel, what am I talking about? It's not a single project to upgrade. It is a flywheel that moves into the cloud and they continue to grow. That's what we're trying to show over time. It allows us to continue to ship our AI faster, monetize all those new use cases, capture those workflows, right? Compounding our momentum with consistent 120%-plus NRR growth at scale. In the cloud, that is what we're seeing. We had this rainbow chart when we went public. I love that we just keep adding years to the rainbow chart and it keeps going.
- 中性 93% 迁云的 DC 客户落在 Premium/Enterprise，迁移时不降档（迁移贡献 ARPU）
What does this look like? What is an illustrative example of customer pricing? It's always hard, so I picked an example. 93% of our data center customers that are upgrading to the cloud are landing in premium or enterprise. And I know this because in some of the models and some of the things we've seen, you can think, oh, well, the data center customer could go backwards moving to cloud standard. That doesn't happen for 93% of those data center customers can mathematically happen, but I thought it was useful to say this. This is -- what did we put 5,000 -- sorry, 5,000 users data center tiers.
- 支持 Jira 与 Confluence 在云端加速；Jira ARR 超 25 亿美元，Confluence 超 15 亿，Service Collection 超 10 亿且增 30%
And that is an incredible set. And people keep asking, well, how do Jira and Confluence keep chugging and must be slowing down, right? And it's not the case. Today, Jira and Confluence are accelerating, as we have said, in the cloud. And it's because of the platform, right? We serve more and more workflows and business processes, which I think people greatly underappreciate, right? With the [indiscernible] collection, people aren't buying point products anymore. They're buying a collection of things that work together that are part of the platform. And that's a massive signal for us, as we've shared, with 1,000 customers, that was a quarter a bit ago, I should say, that was in previous quarter's results and more than 1 million seats in the first six months. So those are some really, really strong numbers, but the point is diversified business, multiple large businesses, all of which are growing very strongly. And again, the collections make it easier to expand.
- 支持 订阅 ARR（DC+Cloud）最近 3 个季度加速——管理层推荐用 ARR 替代收入看增长（图为示意，无数字）
So why is that? Well, ARR normalizes the effect of ASC 606, that revenue recognition and helps everyone, I think, much better understand the overall health of our business. So if you take our subscription business, so data center and cloud, put it together, do a lot of great mathematics that our finance team is able to do, you see a very different pattern. Here is our subscription ARR for the last 7 quarters. What you can see in the last 3 quarters is a pretty good pattern of acceleration of ARR at the overall level of the business. One of many indicators, but I think a really strong one showing how strongly we feel about how our business is operating and trying to explain the dynamics between all of these different things. I'm sure we'll have time for questions on that. Last one, I wanted to talk about one more thing before Brian comes on.
- 口径不齐 DC 侧前移/拉动说明 FY26 部分增长有一次性成分（DC 许可提前确认）
We have a conversation, yes, we're good at that, et cetera. And lastly, data center customer purchasing patterns have continued to emerge and accelerate as we move to cloud. Again, all these are good things. We saw that play out in a pretty big way last quarter, as I know a number of you heard about, with a lot of pull-forward activity into FY '26 from FY '27, which resulted in greater licensing revenue in a DC in '26, far more than we had expected, which is a good thing. In general, timing dynamic creates a problem. So we continue to see strong cloud growth rate, great for our customers, great for our business and data center retention, but the greater term revenue in '26 in data center, we now expect negative data center revenue growth in '27. We've pulled all that forward, which is a great thing for us.
- 支持 $1M+ 客户超 600 个，4 年 6 倍、过去一年 +39%；$3M 客户 4 年 10 倍、+54%；留存 >99%
Well, we talk about having 350,000 paying customers. We know that that's the bottom of this kind of pyramid illustration. How about at the top end? So our $1 million customer base, which I believe we said a quarter and a bit ago, had passed 600 customers spending more than $1 million a year. That has grown sixfold in the last four years. In the last year alone, it grew at 39% year-on-year, that category. These customers have more than a 99% retention year- on-year, and they are almost all expanding. If we look at our $3 million cohort over the last four years, when we talk about the enterprise investments we made on the technology side, on Brian's side, that $3 million cohort has grown tenfold in the last four years, 54% year-on-year in the very last year alone. And we have quite a number now of those $10 million-plus ARR customers.
- 口径不齐 数据中心提价导致 Q3 拉动；准备上云客户扩张放缓；FY27 收入机械性低谷、FY28 再加速，ARR>20%
Yes, that's right. And that's a big part of the reason why we disclosed ARR recently, right, to really just showcase this overall business in the right light. It removes a lot of the noise as it relates to data centers. So for those maybe not following us closely, what's happening on the data center side is that in September of 2025, we announced our data center end of life. And what that did was create a couple of dynamics. One is there's ASC 606 rev rec that got pulled up from future periods into FY '26. That's one dynamic that we saw. Q3, the quarter -- our most recent quarter is our largest installed base in terms of expiry base for data centers. We had a pricing change going to effect in March of this year. And what we saw was that for customers that know that the migration to cloud is going to be a multiyear journey because of all the custom builds, all the change management and the very large user base that they need to make sure they migrate in the right way to preserve business continuity, we saw them pull up their business, right? They expanded and said, look, we're going to continue to commit to Atlassian. We're expanding, and we want to take advantage in the quarter. And so we saw a lot of that revenue also get pulled up into the period. The other dynamic that we saw then is for customers that are actually getting ready and active to move to the cloud, they're going to see much more muted expansion, right, which makes sense. As they move to the cloud, they will continue to expand there. And so those are some of the dynamics, and that's why we wanted to show that in FY '27, we'll see a revenue trough, a mechanical sort of revenue trough in FY '27 with reacceleration out in FY '28, but the area to really point to is ARR growing north of 20% and reaccelerating as well.
- 支持 ARR 增长 20%+，cloud NRR 超 120%；Q3 超预期来自 TWC 与 Jira 独立 seats
Yes. I mean I think, again, you can see it in the performance already, right? We talked about -- we introduced ARR at the Investor Forum growing north of 20%. And if you look at the NRR on our cloud basis as well, north of 120%. So incredibly healthy. We called out the outperformance in Q3 due to Teamwork Collection as well as continued growth and expansion in Jira stand-alone seats. And on the AI side, a couple of the areas that we had shared was the fact that our Teamwork Collection customers, they're using Rovo credits at 2x the rate and deploying 2x the number of agents as those that aren't yet deploying Rovo. Our Rovo customers are growing and using their AI credits 20% month-over- month. And Koji, you asked a question about maybe some stats that we haven't necessarily emphasized as much. If you think about the platform, I talk a lot about Teamwork Collection and then multiple platforms as well, but again, open by nature, right? There's customers certainly that use their own agents MCP into the Atlassian platform. And that, again, compounds back to the Teamwork Graph, and we're seeing MCP users grow at 2x the rate. The number of objects being written to Teamwork Graph right now is growing 100% month-over-month. This is through via MCP, right? So as customers continue to develop their agents, as customers continue to use Rovo agents on top of our platform out of the box, we're seeing incredible adoption across the AI sector. And I would also mention that those customers that are using MCP, their MRR, it's early days, but MRR is growing 2x the rate. And these aren't just existing customers, these are customers that have been both with us for a very long time and also what I would consider as AI native customers, right? If you think about the Forbes AI 50, 2/3 of them are Atlassian customers.
- 口径不齐 DC 迁移客户 93% 进入高版本；cloud 中加人更容易，带来强劲用户增长（迁移相关）
When we think about migrations, we've obviously had a lot of migration success, and we've learned from the end of life of our server product several years ago. And we continue to see that those migrations tick up in the more recent years. More importantly, you're seeing those data center customers when they migrate to cloud, they're migrating to our cloud premium and cloud enterprise SKUs. So 93% of those data center customers, when they migrate, they migrate to the higher-level editions. And then actually, what's more important, once they land in cloud, we have significantly more products to cross-sell them to. We have higher-level additions, which we can kind of upgrade them to over time or upsell them to over time. And then it's significantly easier to add users in the cloud, right? I don't have to ping my IT admin and say, "Hey, can you add Gregg into my instance?" I can actually with little friction, click and add, Gregg immediately, bring in more users. And again, you get the benefit of the whole platform, underlying teamwork graph. So you get this compounding effect.
- 反对 FY27 DC 负增长，拖累总收入增长（606 高基数）
Yes. So as I mentioned earlier, with the announcement back in September when we announced the end of life of our data center product, I think that did a couple of things, right? Like that drew a line in the sand for those data center customers of a time line of when they need to migrate over to cloud. And with that, it also introduced ASC 606 changes of revenue recognition on the data center line because now that kind of useful life of the data center product has shortened with that definitive end-of-life date. And with that, that brought forward a lot more upfront term license revenue into '26 is just mechanical math of more upfront term license revenue. So this is all specific to data center again. And therefore, that creates a meaningfully tough comp for '27 as well, again, just math, mechanical math. And so we do now expect data center to have negative growth next year as that business contracts as more customers migrate over to cloud as well. And so that negative data center growth will be a drag on overall revenue growth next year. And so to help bridge investors through that, we thought it was important to introduce subscription ARR for you to understand the health of the overall book of business, right? That normalizes for all the 606 noise. It also helps investors kind of focus less on probably the migration economics in terms of like, hey, how much uplift on cloud is coming from migrations? How much of the headwind is on data center? Now you get a picture of the full book of business on a normalized ARR basis, and then you can understand the health -- overall health of the business better.
- 口径不齐 分析师引用管理层称 cloud 29% 增长来自 seat expansion 和 cross-sell，专家回答只谈 land-and-expand 和平台整合，没有给出 seat 数据
Atlassian's cloud revenue accelerated 29% in Q3 FY26, driven by seat expansion and cross-sell rather than migrations, according to management.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 $1m+ 客户翻倍主要来自老客户扩张与分散合同合并，而非新类别客户
**BR** [00:43:29]: It's definitely customers that have been Atlassian customers for a while, and they're now expanding their usage of it and consolidating their usage of it. The way that Atlassian used to sell, again, was 10s and 20s and 30s here and there, the 300,000 customers aren't unique customers. There's 300,000, maybe there's and Team 1 through Team 200 at Nike that uses, I'll use Nike again, that uses Atlassian before, but they were all separate contracts. Now consolidated those contracts into one, consolidated into one administrative model allows you to get into the… Actually, you're on the docket for someone like a CIO to even recognise now. You're like, "Oh, it's not just several hundred teams spending a few thousand dollars a year, it's now one team spending USD 1m per year, and that's one of my top vendors I need to spend time with you, which gives us the ability to upsell and upgrade from where you were before." Yes, they're absolutely going up. It's existing customers, it's upselling those existing customers and consolidating their usage, all of those.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 JSM 交叉销售目前只针对不到 10% 客户，存量渗透空间大（受销售产能限制）
**BR** [00:30:01]: Yes, I think so, absolutely. I look at a product like Jira Service Management, I won't give you the exact numbers, but out of their 300,000 customers, they've specifically targeted only a small percentage of them currently to be cross-sell candidates, and then they're going to try and win those first. Then as they win more of those, out of the 300,000 customer base, they'll add more of them to go after. That small percentage is less than 10% as an example. They're not even trying to cross-sell Jira Service Management into more than 10% of their customer base currently. They've still got 90% of the customer base they're not even trying to sell their third biggest product into. Then you've got the Loom, you got a whole bunch of other products sitting behind it that are even at a smaller percentage penetration today. There's a lot of room for them to grow for sure, just internally to their own customer base.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 口径不齐 迁移上云目前也带来收入贡献，增长不全是有机
**BR** [00:24:51]: I think that that range is absolutely within the realm of where they could go. There are a few really important tailwinds that they have going for them. Number one, they're closing down their data centre business, which has been what they've been doing for the last few years. As they move customers to the cloud, those customers can adopt new technologies, not only within the products they currently own, but also other products they maybe didn't own before, they can adopt those much faster and much easier within the cloud. Moving those customers from their data centres to the full cloud enables that, and it's also a revenue benefit at the moment. That's another thing. I think the other tailwind they have going for them is just AI itself as they've been one of the… I was shocked. When I came there from ServiceNow, I had no idea how good their AI actually was. They just weren't talking about it at the time, so I did a lot to try and make sure that we changed the business model around, at least in the IT service management world, to talk more about AI than they were, but the technology has always been there.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 支持 使用 Rovo 的客户 ARR 增速是未使用客户的 2 倍以上；CEO 否认是早期采用者偏差
With our $3 million ARR customers growing over 50% year-on-year and our $5 million ARR customers growing over 70%. In the Atlassian system of work, leading enterprises like Google Cloud are using our platform to power the way their teams collaborate. And in AI, Rovo is now used by over 80% of the Fortune 500. Rovo assisted actions grew 50% over the prior quarter and Rovo adopters that is customers using Rovo continue to grow their ARR at more than twice the rate of non-adopters. When we think about what AI is doing for Atlassian, the results speak for themselves. Customers are getting real value from the AI embedded directly into their workflows, and it's translating into direct growth in our business. Customers are committing more deeply to the Atlassian platform because of the value delivered.
- 支持 迁移符合预期，不是 4900 万美元超预期的来源
Thanks, Ryan. No, the philosophy remains -- really remains the same here. And on the cloud revenue drivers here on the $49 million beat, maybe I'll start with the fact that and highlight the fact that the DC migrations to cloud were really in line with expectations and not a contributor to the $49 million beat. It's progressing well and in line with our expectations for FY '26 contributing approximately mid- to high single digits to the cloud growth revenue. So the 2 primary drivers here of the outperformance were really the cross-sell and seat expansion that we were seeing similar to what we saw in the cross sell side, we continue to see strong uptake in our collections across Service Collection and Teamwork Collection. Service Collection revenue growth accelerated in Q4, just a strong signal of the AI adoption that we're seeing and that Mike mentioned a little bit earlier here. In fact, we're seeing over 3x increase in Agentic automation in Service collection as customers are continuing to deploy more agents beyond IT teams in HR and legal and finance and marketing teams. So again, also great to see the recognition here being named a leader in Gartner's Magic Quadrant for conversational AI platforms for employee services. We also continue to see a lot of strong performance on the Teamwork Collection side. A good reminder that Teamwork collection is still the best vehicle for customers to adopt and unleash AI on the Atlassian platform. This is where customers are upgrading to increased Rovo credits, it's giving customers what they want in terms of the predictability as well as the ability to really load balance those tokens across their organizations, and we're seeing that drive our strong ARPU uplift here as well. So really, I think what this shows is that the customers are consolidating their collaboration stack onto the Atlassian platform, recognizing the value that the Teamwork Graph really brings here. And we do see that in the results with Teamwork customers having 4 to 5x more paid seats per instance than stand-alone Jira or Confluence customers.
- 支持 Fortune 500 覆盖 85%，但收入占比仅约 10%；300 万美元+ 客户 164 家、同比 +50%，500 万美元+ 客户 69 家、同比 +70%
And just maybe to add to that, I do want to reiterate what Mike said that it's not an overnight thing. We've got sort of similar questions last quarter about Service Collection when we had announced the $1 billion ARR growing north of 30% year-on-year. Those are quarters and years of investment that we're making here. And if you take a look at the enterprise go-to-market motion that we're continuing to evolve here, we really are still early stages when you think about the fact that we're in 85% of the Fortune 500, but really that only represents roughly 10% or so of our total revenue. There's still a lot of headroom there. And sure some of these stats a little bit earlier, but our $3 million cohort, that's 164 customers growing 50% year-over-year. Our $5 million and greater cohort of customers, that's 69, and that's on 70% year- over-year. But again, on a base of $6.6 billion of ARR, it really just highlights how early we are here in terms of the opportunity.
- 支持 $49M Cloud 超预期全部来自核心 seat 扩张和 Collections 交叉销售，迁移未贡献
* **Cloud** The most important part of the print was the quality of the Cloud upside. The entire ~$49M Cloud beat came from core Jira/Confluence seat expansion and Collections cross-sell; Data Center migrations were in line and did not contribute to the beat. Cloud has now accelerated from 26% to 29% to 31% over the past three quarters, materially weakening the argument that recent upside was just migration, pricing or acquisition noise. * **Guide** The F1Q and FY27 Cloud guides were substantially better than feared, while the headline +13% total revenue guide is depressed by the expected 17% Data Center decline. The 18% Subscription ARR guide nevertheless implies roughly flat organic net-new ARR dollars in FY27, so this becomes the key metric for determining whether mgmt’s second-half prudence is simply conservatism or reflects genuine decel.
来源背景：TMTB 作者转述业绩会与 callback - 中性 callback：迁移对 Cloud 增长仍贡献中高个位数
* **Competition** TEAM is positioning itself as the open context and orchestration layer rather than a closed AI destination. Mgmt highlighted deeper foundation-lab partnerships, integrations with Claude, Cursor and GitHub agents, and record conversions from legacy service providers. The principal debate is whether Teamwork Graph becomes a foundational enterprise data layer or ultimately remains one feature among several large knowledge graphs. * **Callback** The callback reinforced that mgmt wants investors focused on annual Subscription ARR rather than potentially lumpy quarterly movements; migrations should remain a mid-to-high-single-digit Cloud growth contribution even as absolute migration volumes rise. Mgmt also said FY27 consumption billing should begin, but is unlikely to contribute meaningfully to growth, while JSM sales capacity is being expanded now that the offering meets more enterprise requirements.
来源背景：TMTB 作者转述业绩会与 callback - 中性 land 规模小（20-50 人团队），增长主要靠存量 35 万客户内扩张
So we actually typically land quite small. So our land -- we have a very differentiated go-to-market motion where we've land with a relatively small team, sometimes it's 20 or 50-person teams within an organization and then expand from there. So I would say, actually, our lands tend to be quite small. It's a department choosing Jira, choosing Confluence on their own. And so the opportunity is really via expansion. I mentioned we have 350,000 customers already today. So the opportunity is really expanding within those organizations and helping them realize the value of, again, the Atlassian platform that they may not realize.
- 支持 大部分增量不是靠标价，而是 seat 数增加、版本升级、单品→collection
Right. So does that still leave you with optionality on price ? Yeah. Listen, I think when we go back to the core values of driving value for some of the WHO actually large part of the lift that we see across our our business is actually not necessarily driven by those list price that we're talking about here. It is customers opting to expand opportunity set across our customers, increasing their seat count. It's upgrading their additions from premium to enterprise. It's actually going from single SKUs to collections and then multiple collections over time. So I think that again, gives us a lot of permission, both from a pricing perspective, but we also think this is a moment to take share at the more and more customers. Think about how to get value out of AI.
- 口径不齐 DC→cloud 迁移客户三年后支出平均 1.75x（迁移带来的扩张）
We only introduced that over the last year or so. I believe that's going to continue to compound. If you look at also the data center side, right, as we continue to migrate customers over to the cloud that continues to be a multiyear expansion pipeline. You know, when we look at cohorts of customers that have since moved from data center to cloud three years on, they've spent 1.75 X on average more than before. So that makes sense because you've got a whole suite of applications and collections that's available in the cloud. You've got Rovo, you've got A.I., you've got the teamwork graph sort of opportunity to drive more value as customers continue to migrate to. And you brought this up two times now, and I think it's really worth expanding and expounding on this notion seek growth.
- 支持 Confluence ARR 超 $1.5B，年增 25%+
Cannon-Brookes rejects host Harry Stebbings’ framing that agents turn software into mere databases, saying that holds only for “pedestrian” functional tasks (like buying a charging cable) but not for collaboration-heavy enterprise workflows. He says Atlassian operates one of the top three-to-five MCP servers globally, enabling agents to read and write the Teamwork Graph, but argues that if Atlassian’s interfaces stop evolving to solve customer problems, the company loses relevance to that experience. He envisions a mixed future of horizontal agents, vertical deep-task agents, and persistent interfaces coexisting — not one “Uber agent” or single magic button. As evidence interfaces still carry commercial value, he cites Confluence generating north of $1.5B ARR, growing over 25% annually at scale — large enough to be a standalone public company — despite writing text seeming commoditizable.
来源背景：Atlassian CEO Mike Cannon-Brookes（摘要转述）
市场预期和股价
判断：市场的担忧从"seat 会消失"切换成了"有机增长有没有在加速"。前一个问题管理层答得很充分，后一个问题 FY27 指引的回答是没有。
卖方在业绩会上反复问 seat：2026 年 2 月 Morgan Stanley 说"市场担心 seat 模型"，4 月 Mizuho 说"市场非常担心 seat 压缩"，管理层每次都用定性说法回应，"如果说有什么，我们看到的是相反的"。8 月 FY27 指引出来后，问题变成了"18% 是不是 seat 在放缓"，Wolfe 算出隐含有机净新增 ARR 持平。卖方观点分化：RBC 认为能持续 20% 以上，空头认为"审慎"可能在掩盖真实放缓。
股价已经把第一个问题的答案反映进去了：周收盘价从 2025 年 2 月的 $314 跌到 2026 年 4 月的 $57 (-82%)，现在 $215，是低点的 3.8 倍。最大的两次跳涨就是 Q3 业绩 + Investor Day 和 Q4 业绩，也就是管理层集中反驳"seats 会消失"的两个时点。
数据：IBKR 周 K 线收盘价，2024-10-21 至 2026-10-19。你提到的"跌 78%、涨 2 倍以上"，按周收盘是跌 82%、涨 3.8 倍。
James, welcome to Atlassian. Mike, you may recall my high level of frustration one quarter ago when after I thought it was a pretty good quarter of acceleration, your shares perceived to go materially lower -- continue to be materially lower. And I don't want to minimize that there's a lot more work to be done. But clearly, this is an impressive result. My question relates to a comment in the shareholder letter that strong seat expansion in Jira was a key driver of the cloud revenue acceleration this quarter. And given that there is so much fear about meaningful seat compression at Atlassian being on the horizon, can you unpack the drivers of the seat growth for us? And secondly, is this a dynamic that you think can be durable?
Yes, Gregg, thanks for that and for calling it out. I hope we've been very consistent on our views in that world. We are not seeing any signal of seat compression from customers. If anything, we are seeing the opposite. We are seeing strong expansion numbers, strong cross-sell numbers between collections, strong usage of AI and strong commitment to the Atlassian platform. Many, many competitive wins, a huge amount of consolidation happening into the Teamwork Collection. So we have a lot of green lights in a lot of different places. Similarly, you can see that NRR maintained north of 120% and even ticked up again for, I think, the third or fourth quarter in a row. We have a lot of reasons why that is. Firstly, I would go to high R&D investment in just great quality software, right? That does matter. It always gets ignored in a lot of context, wrong choice of words, in a lot of quarters. But we build amazing applications that deliver great value for our customers. Secondly, the context we have in the Teamwork Graph and the critical business processes continue with AI to blur the boundaries between teams and between roles. That means our platform and our offering between service teams on one side that connect to finance and HR, marketing and in through ops teams to technology teams with business teams and leadership teams. The same context across all of those teams allows us to expand into those nontechnical roles at an increasing rate, right, which shows why we are getting that seat expansion in different collections, in different areas and just the continues in our business. Fundamentally, customers are opting for more and more workflows on the Atlassian platform.
Congratulations on an amazing -- amazingly strong end to the year. I just wanted to ask about the subscription ARR framework. Obviously, it's kind of newer framework to us. And you again had another great end to year 23% growth for the fiscal year '27 guide is down to 18%. Could you maybe just help us out around the difference between Cloud and data center within that and how much of this is kind of some moderation around new seat expansion, et cetera? And how much of it is around prudence surrounding macro and continued scaling of the enterprise go-to-market engine?
RBC argues Atlassian can sustain 20%+ subscription ARR growth as AI increases software creation and coordination complexity, making Jira/Confluence more valuable rather than less. The key differentiator is the Teamwork Graph, which spans 200B+ objects and can act as a control plane for agents; RBC cites early users of AI coding tools showing ~5% more Jira tasks, higher MAU and faster seat expansion. Collections/Rovo adoption and improving enterprise GTM provide additional upside, while the end of the multi-year platform R&D cycle should support meaningful margin expansion.
来源背景：RBC 卖方（TMTB 转述）
**The bear case** is that the quarter was excellent, but the FY27 framework still signals a meaningful underlying slowdown. Subscription ARR falls from 23% to 18%, which implies approximately flat organic net-new ARR dollars, and total revenue slows from 26% in FY26 to 13% in FY27. Cloud guidance of 25.5% is strong, but it still embeds deceleration as TEAM laps two quarters of acceleration, while Data Center declines 17%. Bears will argue that mgmt’s “prudence” could be covering real moderation in seat expansion or cross-sell as the year progresses.
来源背景：TMTB 作者转述业绩会与 callback
展开全部 34 条原文 (按时间)
- 中性 重申到 FY27 收入 CAGR 超 20%
Yes. Thanks, Mike. There's nothing in the macro that changes our thinking on our commitments that we made at the Investor Day. In general, I would say, we remain confident and on track to the plans we laid out at that time. From a revenue perspective, we do expect to tap into large market opportunities to drive healthy revenue growth across Cloud and Data Center through our strategies that Mike talked about earlier around enterprise, AI and System of Work. All of that culminates in our ability to drive revenue growth in excess of a 20% CAGR through FY '27. Rovo net helps that, and we don't see anything in the macro right now that would prevent us from accomplishing that.
- 中性 SAM 670 亿美元、run rate 50 亿美元；存量客户里有 180 亿美元机会，其中 140 亿在企业客户
Yes. That's interesting because the way we look at it, the SAM for the things we build is about $67 billion, and our annual run rate is right now at $5 billion. So there is definitively a lot more that we can go grab. And like I said earlier, the 300,000 customers, companies we serve, there's millions of companies in the world and our products that really is able to serve both technical and non-technical teams. So we really have a lot of upside. But very quantitatively, if you want to look at it, I think our unique land and expand model, the business model helps us be very efficient with how we go to market. So when we went back and analyzed just within the 300,000 customers, let's say we didn't acquire a single new customer, what happens? What's the opportunity there? There's an $18 billion opportunity just within existing customers even if we didn't go out and acquire a single new customer and $14 billion of that is within enterprises. This is why adding hyperefficient enterprise go-to-market motion for us is a clear winner, right, that we can go and access a lot more within our existing base. Having said that, new customer growth continues to be healthy for us, and we expect it to continue to grow and expand simply because of the wide variety of customers that we serve and the power of having a platform that lives in the future.
- 口径不齐 分析师问 FY27 目标（收入 CAGR 20%+、利润率 25%+），Lam 只重申利润率，没有回应收入 CAGR
I don't know if I have any additional comments to say. As you mentioned, we've committed to 25% plus non-GAAP operating margins by FY '27. We've made fantastic progress over the past 2 years, in particular, in terms of margin expansion this year, in particular, I'm really proud of the kind of the efforts across the company. Look, I think we've talked at least at our Investor Day about in the years to come, you'll see R&D as a percentage of revenue moderate. And at the same time, sales and marketing will modestly tick up in the years to come as we invest more against the enterprise opportunity.
- 中性 Q4 指引原本假设宏观和 GTM 调整会对 paid seat expansion 有负面影响，结果没发生
So overall, a very strong quarter of progress and execution in the enterprise portion of our business, which drove the bookings and billings performance you see in the results. Now if you go down to revenue, that was also better than our expectations. As you saw in the numbers, we had outperformance on Cloud and marketplace and in-line performance on Data Center. On Cloud revenue guidance coming into the quarter, we'd assume some negative impact on paid seat expansion rates and cloud migrations from macro uncertainty and our Enterprise go-to-market evolution. And to your point, neither one of those were fully materialized. We continue to benefit from relatively stable macro and business trends from the quarter. In terms of our Cloud performance in Q4, paid seat expansion, cross- sell and migrations were the primary drivers of the better-than-expected results with in-line performance across the other Cloud drivers. Notably, the rate of paid seat expansion in SMB and enterprise continued to demonstrate quarter-to- quarter stability.
- 中性 卖方（Morgan Stanley）指出市场担心 code gen 削弱开发者角色，与管理层看法相反
Congratulations on a really strong end to the year. Mike, reading the shareholder letter and kind of listening to what you've been describing going on Atlassian, is a little bit of an odd that we hear in the marketplace, frankly. There's a lot of concern around cogeneration tools and the changing role of a developer, which you seem to see as a positive versus the market seeing as more of a negative. Maybe you can walk us through what you guys are seeing in the business today in terms of seed expansion with developers or more broadly? And what do you think is going to transpire on a go-forward basis in terms of how these innovations are going to affect that core business from Atlassian going forward?
- 中性 CFO 重申 FY24-FY27 收入 CAGR 20%
Yes. Thanks, Kash. This is Joe. Great question, and I'll start and then pass it to Mike for more detail. So the first point I'd make is we have not changed our approach from last year with respect to our guidance for Q1 and FY '26. As we discussed last quarter, our guidance approach for FY '26 is consistent with the approach we took in FY '25, which is we are taking what I believe to be a conservative and risk-adjusted approach to account for macroeconomic uncertainty and potential business disruption from the evolution of our enterprise go-to-market sales motion because we believe those risks are still very relevant to the current operating environment. In addition, I'd say there's no change to our outlook that we will be able to deliver 20% compounded annual growth from FY '24 through FY '27. And we believe with 20% growth in FY '25, we have laid a strong foundation for achieving that and are on track to that goal. Our confidence in that remains rooted in the significant market opportunities we have in front of us. Our strategy to capture them through our investments in Enterprise AI and System of Work. And the strong momentum we've built around those strategies, as Mike laid out in the shareholder letter and in his opening comments. We have multiple levers of growth from paid seat expansion to cross-sell, upsell, pricing and new customer growth. And AI opens the aperture for even more opportunity given the structural position we have with our cloud platform and Teamwork Graph and deep investment in R&D. So taken together, the guidance approach and the opportunity for growth that we have, there's no change to the 3-year outlook we provided at our Investor Day last year, and we remain very optimistic about our ability to grow at a very healthy rate over a sustained period of time.
- 中性 Q1 其他有机驱动略好于或符合预期，对迁移以外的变量继续保守
Keith, this is Joe. I'll go first, and Mike will follow on with additional context. I'd say the key development that we saw in Q1 was that we had significantly stronger-than-expected cloud migrations from data center, which is a great thing for our business. As you know, this has been a big area of investment for us. It's a key strategic priority. It allows us the opportunity to provide more value to customers in ways that we simply can't on data center, and we can offer capabilities such as automation, analytics and AI, and so basically, it's definitely the most valuable and secure experience we can offer to our customers. So it's a great thing for the business. Cloud migrations also, however, have an impact on the timing of revenue recognition because cloud revenue is recognized ratably and data center has a combination of upfront recognition and some ratable. And then lastly, the move to cloud also impacts marketplace revenue because we have a lower take rate on cloud app sales than we do on data center apps. All other organic growth drivers in our business in Q1 were either slightly better or in line with our expectations. And we've maintained that same guidance approach on those factors that we held 3 months ago. So we continue to hold a very conservative and risk-adjusted outlook for all the other variables in the growth equation outside of migrations. So with that Q1 performance and momentum, we've adjusted our full year outlook for a greater cloud migration forecast. And we've left all other organic driver assumptions in place. And because of those revenue recognition timing differences between cloud and data center and the impact to Marketplace, this drives the 0.5 point decline in our organic revenue growth outlook for the rest of the year. And so that's the math underneath the guidance that we're giving for the full year on the organic part of the business. Mike?
- 中性 CFO 重申到 FY27 收入 CAGR 20%+、云收入健康且加速
Yes. Sure, Raimo. There's no change to our midterm outlook calling for 20% plus compounded annual revenue growth through FY '27. I'd say the same thing for our 25% plus non-GAAP operating margin commitment in FY '27. We remain confident in our ability to deliver healthy and accelerating Cloud revenue growth as we expand operating margin over time. And I'd also highlight that with respect to our short-term guidance for FY '26, we do continue to take a conservative and risk-adjusted approach. So that's the way to think about that.
- 中性 分析师指出市场担心 seat 模型，客户可能仍偏好 seat 定价
Yes. This is Sanjit Singh for Keith Weiss. Congrats on the $1 billion quarter in Cloud and the progress with Teamwork Collection. Really nice to see. I want to follow up on Rob's question in terms of kind of the reality on the ground in terms of what the market is sort of asking and calling for versus what customers want and this is specifically around pricing. And I'd love to hear your take on where does pricing go to and evolve over the next 1 to 3 years. Market seems concerned on seat-based model, seat-based pricing. Customers probably like seat-based pricing. And so what -- where do you think this all shakes out in terms of where we're headed in terms of the pricing monetization story for Atlassian?
- 支持 首个 10 亿美元季度、600 家年付 100 万美元以上客户、RPO 增长超过 40%
Sure. So I would say, firstly, we're not seeing a reduction in the number of seats, right? The conversations that we're having with customers are not around that. We're not seeing that in the conversation. We're not seeing that at the time of renewal with our customers. Instead, the conversations that we're having with our customers are very different. I'll give you an example. Just earlier today, I spoke with a large U.S.-based consumer tech company that was talking to us about a long 7-year term agreement that they wanted to enter into us, and this is a company that has over 100,000 seats with us both in the developer and the IT space. We firmly believe that having the ability for developers and nondevelopers to create software leads to more software. And same as what Aaron said, this means that there will be, one, the hiring of more developers and the creation of more software. And more software leads to more projects. What we have found, in fact, is that we see -- and for companies that are utilizing these cogeneration tools, there actually is a 5% increase for us in seats. And a 5% increase in fees obviously leads to expansion with our customers. And I know that's contrary to the headlines that you just gave there, but this is having a tailwind for us rather than a headwind. And that is what is leading to the headlines that we made around $1 billion -- our first $1 billion quarter in Q2, the 600 customers that are now paying us over $1 million and an RPO growth that is north of 40%. If we were facing all of these headwinds, we wouldn't be able to deliver numbers consistently like this quarter-over-quarter. So if anything, like I said at the beginning, AI is actually a tailwind rather than a headwind for us.
- 中性 分析师称市场普遍担心 Atlassian 的 seat 会明显压缩
James, welcome to Atlassian. Mike, you may recall my high level of frustration one quarter ago when after I thought it was a pretty good quarter of acceleration, your shares perceived to go materially lower -- continue to be materially lower. And I don't want to minimize that there's a lot more work to be done. But clearly, this is an impressive result. My question relates to a comment in the shareholder letter that strong seat expansion in Jira was a key driver of the cloud revenue acceleration this quarter. And given that there is so much fear about meaningful seat compression at Atlassian being on the horizon, can you unpack the drivers of the seat growth for us? And secondly, is this a dynamic that you think can be durable?
- 中性 FY27 指引留到 8 月；将披露历史 subscription ARR 以平滑 DC 确认节奏
Arsenije, thanks for the question. As it relates to FY '27, right now, it's too early to discuss any guide at this point, we'll, of course, provide that guidance out in August, our Q4 earnings. But as it relates to ARR, as I mentioned a little bit earlier, right, we're seeing that lumpiness in the revenue recognition in the year on the data center side. And next week, we're going to be able to share some of the historical subscription ARR across the overall business to help really smooth out those timing effects that we talked about. And I think that will give folks a better understanding of the underlying strength in the business. But maybe just as a reminder, too, for the data center end of life announcement, right, we began to recognize great upfront term license revenue and that results in greater upfront revenue recognition in the period, but there's a corresponding drawdown in RPO and in particular, CRPO as well. So it's worth sharing then that when you normalize for the impact of ASC 606 here, our RPO would have been north of 40% in the quarter, in Q3. And our CRPO would have been north of 30% year-over-year in Q3, much more in line with some of the recent trends that we're seeing, and again, underscores the strength in the backlog that we're continuing to build.
- 支持 多头认为本季回应了 Cloud 持续性和 AI seat 风险两大去估值原因
**Bulls** will argue this quarter directly attacked the two biggest reasons TEAM had de-rated: Cloud durability and AI seat risk. Cloud accelerated to 28.6% y/y, well ahead of Street, and the drivers were not just migrations. Mgmt and callbacks pointed to Jira paid seat expansion and cross-sell into Teamwork Collection and Service Collection, with NRR still north of 120% and improving. That matters because the bear case had been that AI coding tools would reduce developer headcount and pressure TEAM’s seat-based model. Instead, mgmt said it is seeing no signal of seat compression, and the quarter showed both standalone Jira seat growth and TWC adoption.
来源背景：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访 - 反对 管理层预期 FY27 总收入增速见底、DC 收入负增长（Ascend 导致 DC 收入前移到 FY26），FY28 再显著加速；放弃 3 年 20%+ CAGR 目标
So I tried to explain this, both overcoming a bulls***, and I'll give you some illustrative charts. I thought I'd say all that so you were listening, I see you were writing before I showed you the charts. So this is an illustrative revenue growth. Number one, we're expecting a trough in total revenue growth in FY '27. Number two, we expect that to significantly reaccelerate in FY '28 as we lap that data center effect going through. You can see why that 1-year acceleration is a credit to the engineering team, creates a challenge for the finance team. And given these dynamics, the 3-year 20% CAGR that we set 2 years ago is no longer a relevant target to anchor on. What is relevant and a much better measure of the strength of our business is ARR.
- 反对 此前 3 年 20% 收入 CAGR 目标不再适用
Number one, we're expecting a trough in total revenue growth in FY '27. Number two, we expect that to significantly reaccelerate in FY '28 as we lap that data center effect going through. You can see why that 1-year acceleration is a credit to the engineering team, creates a challenge for the finance team. And given these dynamics, the 3-year 20% CAGR that we set 2 years ago is no longer a relevant target to anchor on. What is relevant and a much better measure of the strength of our business is ARR. Again, this is illustrative, don't get out your rulers. So why is that? Well, ARR normalizes the effect of ASC 606, that revenue recognition and helps everyone, I think, much better understand the overall health of our business. So if you take our subscription business, so data center and cloud, put it together, do a lot of great mathematics that our finance team is able to do, you see a very different pattern. Here is our subscription ARR for the last 7 quarters. What you can see in the last 3 quarters is a pretty good pattern of acceleration of ARR at the overall level of the business.
- 中性 管理层指引 FY27 收入低谷、FY28 再加速，ARR 增长 20%+ 并再加速
Yes, that's right. And that's a big part of the reason why we disclosed ARR recently, right, to really just showcase this overall business in the right light. It removes a lot of the noise as it relates to data centers. So for those maybe not following us closely, what's happening on the data center side is that in September of 2025, we announced our data center end of life. And what that did was create a couple of dynamics. One is there's ASC 606 rev rec that got pulled up from future periods into FY '26. That's one dynamic that we saw. Q3, the quarter -- our most recent quarter is our largest installed base in terms of expiry base for data centers. We had a pricing change going to effect in March of this year. And what we saw was that for customers that know that the migration to cloud is going to be a multiyear journey because of all the custom builds, all the change management and the very large user base that they need to make sure they migrate in the right way to preserve business continuity, we saw them pull up their business, right? They expanded and said, look, we're going to continue to commit to Atlassian. We're expanding, and we want to take advantage in the quarter. And so we saw a lot of that revenue also get pulled up into the period. The other dynamic that we saw then is for customers that are actually getting ready and active to move to the cloud, they're going to see much more muted expansion, right, which makes sense. As they move to the cloud, they will continue to expand there. And so those are some of the dynamics, and that's why we wanted to show that in FY '27, we'll see a revenue trough, a mechanical sort of revenue trough in FY '27 with reacceleration out in FY '28, but the area to really point to is ARR growing north of 20% and reaccelerating as well.
- 中性 分析师追问 AI 颠覆担忧和估值，CFO 只回答'看业绩'
Mike, your founder, co-founder has been on a lot of podcasts. And I'll go on a walk and listen to it, and he keeps talking about 20% growth, 20-plus percent growth, like we're in this, and we -- I just don't see the AI destruction that Wall Street keeps talking about, yet your multiple keeps fading. So this -- I guess, the question I get is this conviction level in this durable growth engine, it doesn't feel like you guys are giving up on that yet. Investors are questioning right now, maybe it's the excitement of memory and the AI party going on in infrastructure and everyone's left the software complex, which I think is partly true. But what -- I mean, if you keep putting up these type of growth rates with this type of margin at this multiple, I don't know how that continues. Something's got to give, and...
- 支持 投资者对 AI 的担忧已由逆风转为潜在增长驱动
Wells Fargo highlighted TEAM as one of its preferred 2H software ideas, citing improving seat growth, accelerating Teamwork Collection adoption, and growing AI-related developer activity. The firm noted 3Q strength was driven by Jira seat growth and Teamwork Collection upgrades, with Teamwork Collection becoming a larger ARPU uplift than initially expected. Wells also sees incremental upside from datacenter migrations, Rovo adoption, and stronger developer hiring trends, arguing investor concerns around AI have largely shifted from being a headwind to a potential growth driver.
来源背景：Wells Fargo 卖方（TMTB 转述） - 中性 Subscription ARR 增速 20%→22%→23% 加速（分析师复述、IR 确认）
And then to be specific about it, so we have 3 quarters of disclosure 3 quarters ago, which was your September quarter, the growth on subscription ARR, I believe, was 20%. And then in December, it was 22% and then this past March was 23%. So we've seen acceleration in that metric.
- 反对 财报前 KeyBanc 下调 FY27 Cloud 的迁移和有机增长假设
KeyBanc lowers its Atlassian PT to $115 from $130 (maintains Overweight) after fine-tuning its model on feedback from investors, management, and partners. FY27 Cloud estimates come down on lower migration and organic growth assumptions, partially offset by slightly higher Data Center numbers. The firm expects management to guide FY27 conservatively out of the gate, but views that reset as a clearing event that improves visibility into near-term model mechanics rather than a thesis change.
来源背景：KeyBanc 卖方；M-sci 数据（TMTB 转述） - 中性 M-sci：6 月趋势偏软，但新 logo 环比小幅改善
Hearing M-sci saying customer logo adds improved slightly q/q despite softer June trends
来源背景：KeyBanc 卖方；M-sci 数据（TMTB 转述） - 中性 BTIG 预期 FY27 Cloud 指引 23%，低于 24% 共识
BTIG’s partner checks suggest Atlassian finished the year strong, and the firm models a 23% FY27 Cloud growth guide vs. 24% consensus with Data Center cut to a 13% decline on Q3 pull-forward — implying 12% total revenue guidance vs. 14% consensus. But BTIG argues most investors already expect below-consensus guides, skewing risk to the F4Q print itself. With ~23% subscription ARR growth, Rovo momentum, and shares at <30x FY28 GAAP EV/FCF, BTIG reiterates Buy.
来源背景：BTIG 卖方（TMTB 转述） - 支持 专家认为约 30% 增长区间在可能范围内
It's been really, really good. I think they're one of the best in terms of SaaS vendors and the work they've done with the AI technology. That's another tailwind they have. I think the other one is that they've invested quite a bit in the enterprise sales force over the last, call it, 5-7 years. They didn't have a sales force, an enterprise sales force, up until then, and in the last two years, they've doubled down on that model. Their 300,000 customers which represent most of the biggest companies in the world who were just buying 10 here, 20 here, 100 there are now being consolidated to bigger organisations and getting the benefit of having a single platform. Just that massive customer base that they have, now they have a sales force to go, I'll say, exploit it to a degree. I think that 30% where in the range where they're currently growing is absolutely within the realm of possibilities for them.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 反对 FY27 subscription ARR 指引 18%，低于 FY26 的 23%
Congratulations on an amazing -- amazingly strong end to the year. I just wanted to ask about the subscription ARR framework. Obviously, it's kind of newer framework to us. And you again had another great end to year 23% growth for the fiscal year '27 guide is down to 18%. Could you maybe just help us out around the difference between Cloud and data center within that and how much of this is kind of some moderation around new seat expansion, et cetera? And how much of it is around prudence surrounding macro and continued scaling of the enterprise go-to-market engine?
- 反对 分析师：FY27 ARR 指引隐含有机净新增 ARR 大致持平
Overnight success, 3 years in the making. I guess maybe just on the initial ARR guide, James, I think it implies roughly flat organic net new ARR in fiscal '27. And you said you're reflecting the same level of prudence that you have historic -- last historically in this outlook. Is there anything different exiting fiscal '26 that increases the confidence in those numbers given some of those dynamics that you guys just went over? Or just maybe just help level set that a little bit?
- 口径不齐 CFO 把 18% 归因于审慎（宏观、财政、地缘）、DX 收购 1pt 的高基数、H2 加更多保守，没回应 seat 扩张是否放缓
Thanks, Adam, for the question, and welcome to the coverage. First, I would highlight that we're really pleased to see the strength that we saw in the quarter, printing ARR growth, sub-ARR growth of 23% year-over-year. When it comes to the guide, we'll continue to take a thoughtful and prudent approach here, really balancing the momentum that we're seeing in the business, the demand signals that we're getting from our customers with the uncertainty as you look out in the macro environment, fiscal policy and geopolitical landscape. And as you mentioned, this is the first time we'll be issuing ARR guidance for the full fiscal year. We're also still early in the fiscal year here. So taking a bit more of a prudent approach in that second half of FY '27. It's also worth noting that in FY '26, the ARR year-over-year growth rate benefited by approximately 1 point from the acquisition of DX at the close of November in 2025. That's our fiscal Q2. And so we'll be lapping that DX acquisition and the healthy acceleration that we saw in the second half of FY '26. And of course, we're going to still continue to scale our enterprise go-to-market sales motion as we unlock the opportunities that we see in front of us. So all I feel really good about the state of the business here. And coming off a quarter where RPO landed at 44% year-on-year NRR was north of 120%. So you're just taking a balanced and prudent approach to the guide.
- 支持 F1Q Cloud 指引 28.5% 远高于买方 22-23% 预期
**TEAM +32%:** A very good looking print, with the biggest positive coming from Cloud. F4Q revenue beat Street by ~7%, while Cloud growth of 31% landed ~1.5–2pts above buyside expectations. F1Q better as well: revenue guidance came in above Street and 28.5% Cloud growth was well ahead of ~22–23% buyside bogeys. RPO +44%. FY27 revenue growth guidance of 13% is slightly below Street, but should be looked through as secondary. “Up to $250M” founder repurchase plan. Not much to pick at here
来源背景：TMTB 作者 - 口径不齐 FY27 Subscription ARR 18% 指引意味有机净新增 ARR 大致持平，是判断真减速与否的关键
The entire ~$49M Cloud beat came from core Jira/Confluence seat expansion and Collections cross-sell; Data Center migrations were in line and did not contribute to the beat. Cloud has now accelerated from 26% to 29% to 31% over the past three quarters, materially weakening the argument that recent upside was just migration, pricing or acquisition noise. * **Guide** The F1Q and FY27 Cloud guides were substantially better than feared, while the headline +13% total revenue guide is depressed by the expected 17% Data Center decline. The 18% Subscription ARR guide nevertheless implies roughly flat organic net-new ARR dollars in FY27, so this becomes the key metric for determining whether mgmt’s second-half prudence is simply conservatism or reflects genuine decel. Mgmt cited early-year visibility, tougher second-half comparisons, the one-point FY26 DX benefit and macro uncertainty * **Enterprise** TEAM signed its largest enterprise deal ever and posted record activity across the $1M+, $3M+ and $5M+ deal cohorts. Customers above $3M of ARR grew more than 50% y/y and those above $5M grew more than 70%, while RPO accelerated to 44%, cRPO to 27% and NRR remained above 120%. Mgmt characterized this as the result of several years of product, compliance, sales and customer-success investment rather than a one-quarter unlock.
来源背景：TMTB 作者转述业绩会与 callback - 反对 空头：管理层'审慎'可能掩盖 seat 扩张或交叉销售的真实放缓
**The bear case** is that the quarter was excellent, but the FY27 framework still signals a meaningful underlying slowdown. Subscription ARR falls from 23% to 18%, which implies approximately flat organic net-new ARR dollars, and total revenue slows from 26% in FY26 to 13% in FY27. Cloud guidance of 25.5% is strong, but it still embeds deceleration as TEAM laps two quarters of acceleration, while Data Center declines 17%. Bears will argue that mgmt’s “prudence” could be covering real moderation in seat expansion or cross-sell as the year progresses.
来源背景：TMTB 作者转述业绩会与 callback - 中性 分析师：公司一直把付费 seat 扩张和交叉销售作为主要增长驱动
KeyBanc Capital Markets Inc., Research Division All right. We can go ahead and get started since we're starting a little late. Most of you know me. My name is Jason Celino. I'm one of the software analysts here at KeyBanc. Great pleasure welcoming back Martin Lam, Head of IR for Atlassian. First, big congrats on the quarter. Excellent Q4 full year results. There seems to be some significant momentum in the business right now. It's great to see the numbers reflect that. So on the cloud side, specifically, you've seen 5 straight quarters of acceleration. You're consistently pointing to paid seat expansion and cross-sells, the primary drivers of the growth. Maybe can you just unpack those elements and explain, like, what's driving this consistency?
- 反对 FY27 Subscription ARR 指引 18%，较 Q4 23% 减速（管理层称偏审慎）
Yes. So we introduced subscription ARR back in May at our Investor Forum for the first time to help investors understand, kind of, the underlying health and strength of the business and our subscription base. We are, as I mentioned earlier, going through a cloud transition as we sunset our data center offering and migrate customers to the cloud in the coming years. And so to help normalize for some of the 606 noise that I mentioned earlier because we now have greater upfront term license revenue recognition on the sale of data center subscriptions and cut through all that accounting noise and timing noise, we've introduced subscription ARR, which, again, smooth things out and helps give you a better read on the underlying strength and momentum in the business. That continues to track incredibly well. We are guiding to that for the first time. We came off a quarter where we grew subscription ARR 23% year-over-year. And we're guiding initially for fiscal '27 to end fiscal '27 at 18% growth year-over-year. And so that's our initial guide. Obviously, we want to take a prudent approach with that guide. It is a new muscle for us. But again, I think it helps investors understand the entire subscription base, so both data center and cloud, cut through the ASC 606 noise and help you identify or I guess, cut through the noise of the migrations of people going from data center to cloud because I think historically, investors have had outsized focus on, say, how much of the cloud revenue growth is coming from migrations. Subscription ARR helps normalize for that because if someone moves from data center to cloud, you kind of just see that all within the overall ARR results.
- 支持 RBC 认为可持续 20%+ subscription ARR 增长
RBC argues Atlassian can sustain 20%+ subscription ARR growth as AI increases software creation and coordination complexity, making Jira/Confluence more valuable rather than less. The key differentiator is the Teamwork Graph, which spans 200B+ objects and can act as a control plane for agents; RBC cites early users of AI coding tools showing ~5% more Jira tasks, higher MAU and faster seat expansion. Collections/Rovo adoption and improving enterprise GTM provide additional upside, while the end of the multi-year platform R&D cycle should support meaningful margin expansion.
来源背景：RBC 卖方（TMTB 转述） - 中性 分析师（买方）问题本身预设生产力提升不带来 seat 增长
**分析师：客户购买x个席位，并为每个席位每月支付y美元。我理解，大语言模型将提高客户的生产力。但在生产力提升并不意味着席位数量增加的情况下，Atlassian如何从中受益？是因为客户通过提高生产力获得了更高的利润率，所以Atlassian能够进一步提价，还是其产品存在更根本的价值？**
来源背景：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团；自称并非 Atlassian 专家 - 口径不齐 Q4 ARR 23% 中约 1pt 来自 DX 收购；FY27 指引审慎（宏观、新指标）
So subscription Emrah was introduced back at our Investor Forum in May really to help Tierpoint sort of reduce some of the noise. Right. It's the best indicator in terms of the momentum that we're seeing in the business right now across across Cloud and DC. 23% year over year in Q4. You know, as we think about what also contribute to that 23% in Q4, there's a point of growth driven by Detaxe acquisition that we did in our fiscal Q2 as part of that. And you know, as we look out into FY 27, number one ARR is a new metric, as you mentioned, that we're introducing for the first time and making sure that we're being prudent around that as well as some of the macro uncertainty, the fiscal policy that I think we wake up to every single day right now. And it's still early in the year in terms of, you know, the growth drivers going into the rest of 27. It's about execution.
Agent 怎么收费：seat 之后
判断：管理层自己在为 seat 之后做准备。最大部分的价值今天仍然按 seat 收，但 credits、用量和 flex 承诺制都已经上线，agent 不按 seat 收。
CEO 说大多数应用按 seat 定价，"这是多数客户想要的买法"；同时在和大客户谈 flex：承诺一个金额，可以在 seats、应用和 collection 之间挪用。CFO 说 AI 变现今天走 TWC 的 ARPU，"以后可能更多转向用量，和 seats 并行"。IR 说用量定价会是"更大的一部分故事"，今年开始执行 credit 上限。对 thesis 的含义是：就算 agent 带来了巨大的工作量，体现在报表上也更可能是 credits 和用量，而不是 seat 数。
We have multiple pricing models. We're trying to be customer-led. We're trying to meet customers where they are. Most of our apps are priced per seat. I still think that's the way most customers want to purchase software. When I talk to customers, that is what they say they want. We have a very generous number of Rovo Credits involved in those per seat accounts. So we have the price per seat. I'm sure you're familiar with that part of our model. And these are sort of illustrative down the bottom.
So it's a different way of thinking about it. And we have a flex-based model that we are working on with customers at the moment that we're actually talking to a whole bunch of customers this week, which is more of the overall sort of commitment-based model, as I would say, is the way to think about it. So as our bigger customers that get well north of $1 million, say, here's what I want to commit to your platform. I want to be able to move it around seats, move it around applications and collections and I want more usage-based certainty. That is another model that we have. Again, we're here for the long term. This is us being customer-led as they ask.
I think different companies are on different arcs of that right now. I don't think it's one size fits all. I think what we're hearing writ large from our customers is, as it relates to AI, is they do want to unlock AI. We're seeing that in our performance in things like Teamwork Collection, where they get 10x more Rovo credits and the growth and opportunity that we see there. But they're looking also for predictability, right? We're hearing a lot right now about token maxing and making sure folks are being really mindful about the tokens that they're spending right now. And so they're looking for predictability. And they're also looking for the ability to load balance those AI credits as well, right? So for any enterprise and organization right now, oftentimes, what we see and what they see is that they'll see hyper users of those credits, and then those that are deployed, but not necessarily using those credits. So it's helpful for them to be able to take a package of credits and tokens and being able to load balance that across their organization as well. Right now, for Atlassian, the best vehicle to monetize AI and the best way for our customers to unlock AI is through Teamwork Collection because when they buy into that collection, they are getting 10x the Rovo credits. And we're seeing a lot of traction with that right now. We see that Teamwork Collection customers are using 2x more credits than those on a stand-alone basis. They're deploying 2x more agents, which then in turn consume more credits as well. And for Rovo customers, we're seeing credit usage grow 20% month-over-month. So a lot of traction on the AI side there. It's being monetized, and we're seeing that today through ARPU in our Teamwork Collection. And I think over time, that may shift over more to consumption, along with seats. And we also introduced our Flex program as well. This is a single wallet for customers, our largest really strategic customers to go all in on the Atlassian platform. That's access to the entire suite across our collections, across all of our applications and being able to unlock all of those credits across all of the different applications and load balance against those applications in terms of the value. And so through Flex, they're able to actually drive the type of value that is most meaningful for them across the entire platform.
Yes. So the primary AI monetization motion or the primary purchasing motion for customers today is upgrading to Teamwork Collection, where you get 10x the amount of Rovo credits. We made the strategic decision about 2 years ago to thread Rovo our AI capabilities throughout the entire platform. And so with each base subscription, you get a certain allotted number of Rovo credits. And then as you approach those limits, then you upgrade to the Teamwork Collection where you get 10x the amount. And that's much more customer-friendly in this moment in time where I think customers want that predictability or value that predictability of. I get 10x the amount of credits across my entire user base at a higher price point, so we realize higher ARPU as a result instead of worrying about Jason is going to chew through x amount of credits and I have to worry about this power user. It's a fungible pool of Rovo credits spread across my entire organization at 10x the amount. So it's much more predictable and customer-friendly, and we're just meeting customers where they're at today. I think over time, consumption-based pricing or usage-based pricing becomes a bigger part of the story. We'll actually begin to enforce Rovo credit limits this year. But again, I think the primary motion you're going to see is customers choosing to adopt Teamwork Collections. So I would continue to pay attention to that because that's probably the primary monetization motion today. Customers are also -- when they adopt Rovo, they, of course, realize significantly more value. And we're seeing Rovo customers grow their ARR at 2x the rate of those that don't adopt Rovo. So actually the concept of delivering value first and foremost to customers, then we'll recoup that value back over time. That's via increased stickiness, faster seat expansion and higher up-tiering to higher value editions. Actually, what's also really interesting is we shared a bunch of MCP stats. So even if customers are using third-party agents, and tapping into the Atlassian platform via MCP, we're seeing significant growth there. So we have over 1 million monthly active users of our MCP server and Teamwork Graph CLI. So again, wanting to tap into the value of the Teamwork Graph, even if you're using a third-party agent. That's delivering significantly more value to the customer. You're seeing that in their output. So they're driving 4x the amount of Jira issued or Jira Work items created via MCP server and Confluence pages created 4x the amount via MCP server. So they're driving significantly more workflows through the Atlassian platform, which is what we want to see. And ultimately, that increased value to customers comes back to us. So exact same number. Those customers utilizing the MCP server are growing their ARR at 2x the rate of those that don't utilize MCP. So I think it highlights the mission criticality of the Atlassian platform, and it shows that it's complementary to even if you're using a third-party agent like Claude agent.
SD: All right. So, big picture. I was going to ask you to define that. But a couple of years out, I definitely think that we're going to see a much evolved Atlassian where they're making their revenue differently than they currently are. So mainly they're a per-seat revenue company. I don't see that staying the same in 2 years or 3 years. I see that they're going to have to change it. The consumption model is pretty much what most SaaS companies are moving towards. We're seeing some of it from Atlassian. There's like 3 aspects of the Atlassian suite that is now consumption-based, which is pretty much assets, automations, and Rovo. So the reason for that is… AW: Shawn, do you think that their migration over the next couple of years to a consumption model will be more profound or more impactful generally than everyone, right? I mean, it seems like with the advent of AI and AI native solutions that whether you're Salesforce, ServiceNow or Atlassian or other, that we're finally seeing acceleration away
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满
展开全部 75 条原文 (按时间)
- 中性 collections 采用统一版本、统一席位数（single seat count），新产品（如 Rovo）直接进入这些用户
I just wanted to add, Keith, from my point of view, maybe slightly more philosophical than Joe. Look, collections are about being as customer first as we can, and we always try to do that at Atlassian. It simplifies how our customers can buy and grow with our offerings. That simplicity, that reduction in friction does result in, in our experience, a long-term growth driver that appears over time as customers increasingly want to buy larger amounts of software from Atlassian, that they want to not buy it in smaller and smaller pieces, if that makes any sense. They want to buy it in large amounts. That enables us to create quite some significant simplicity in the purchasing behavior, but also simplicity in the deployment and the rollout behavior. As Joe mentioned, that simplicity comes from having a single addition, having a single seat count, and any new offerings that are introduced like Rovo, for example, into the Teamwork Collection will just appear for those users, which is a much easier experience for the customer. I believe, over time, that will lead to continued growth, that's part of the reason, but also lead to customer stickiness and customer happiness as more of our functions are exposed to more people within that organization. And it gives us a larger opportunity to go after the knowledge workers that we don't have access to in many of our customers at the moment or who don't have access to our products. So philosophically, I think it's about simplifying things for our customers in terms of their purchasing behavior.
- 口径不齐 AI 用户一个季度内从 100 万增到 150 万（AI 用户数，不是 seats）
So we are not asking people to go buy a new product, learn a new thing in order to use AI. It very much plays into the land and expand model where we're very focused on usage and adoption first and being patient with monetization. So we've included Rovo with every single addition of our products, and we've seen phenomenal adoption of it. So in fact, just within a quarter, we went from 1 million users of AI on our products to 1.5 million. So we've managed to increase the usage and adoption by 50% just over the last quarter. Business-wise, why this is exciting for Atlassian is that AI as a technology wind, as a technology trend basically means that more software gets created. So earlier, you needed to be a highly technical person writing code in order to create software, whereas now that same power of creation is available to everybody.
- 中性 消费计价是在 per-seat 之外获取价值的方式
How do you manage the cost of this? There'll be all these new problems where people have to come together to solve these new problems. And I think that's a good problem for Atlassian because that's what our products serve. I also mentioned earlier that we've introduced consumption-based pricing elements within our products, and that's a way to kind of capture value, not just through a per seat basis. Earlier, I talked about 50% of our users are nontechnical in nature. So I think a lot of people are certainly focused on the developer side, but oftentimes overlook the fact that we serve -- half our user base is nontechnical in nature. I also think like over time, within this user base, you're going to have to see, again, more and more people come together, how do you solve new problems.
- 中性 企业销售要转向价值卖法，而不是只看客户需要多少 seats/license
Yes. So Atlassian is in a unique position because if you were to look at product like growth in a dictionary, I think Atlassian's name would be beside it. So we have a high-touch motion. We have a low-touch motion, and then we have that word, I think everybody would love to have, the no-touch motion. And the move forward plan is we will continue to have all 3 motions. And we have 85% of the Fortune 500 customers are customers of Atlassian today. They make up approximately 10% of our revenue. So there is an opportunity for us to do more, obviously, with those customers. We'll continue to invest in that no-touch motion from an R&D perspective. And then from the, let's say, the enterprise motion, we have today, north of 500 customers that pay us $1 million or more a year. We plan to, one, hire more sellers. Currently, we have approximately 400 sellers only. So when you think of our market cap and the size that we are, we have 400 sellers, right? So we have an incredible opportunity ahead of us. We are focused on getting closer to our customers. And the best way to do that, obviously, is to hire more people. So as we move into FY '26, we will see an increase in our spend on sales and marketing. We will also be focused on value engineering, which means that we will be pulling on the value that our solutions are driving for our customers. I say that because that is in one of the ways that enterprise sales organizations work by really pulling out the value that the solutions are driving for customers in terms of savings, et cetera, et cetera, rather than looking at purely how many seats, licenses, et cetera, does the customer need. And then at the same time, we're going to be working very, very closely with the ecosystem, which is obviously a huge scaling engine for us as an organization. Since I joined Atlassian, I've been super impressed with the ecosystem itself, how large it is. Such a significant portion of our business comes from the ecosystem, which is great, and how we need to invest further into the ecosystem so that they can continue to scale our business as well.
- 中性 开发者占 seats 的很大一部分
Yes. Good question. So for us, we see this as actually really being a tailwind and something that's going to help our business. So obviously, developers are incredibly important to us and are a large portion of the seats, overall seats for us. Now when you think of developers, obviously, a portion of their time is spent literally hands on keyboards in terms of writing code. But actually, only 1/3 of their time is really spent hands on keyboard. The rest of their time is spent collaborating with team members and managing the projects. And they will be doing that predominantly via our tools, Jira, Confluence, et cetera, which is great for us. Now as new tools like you highlighted, are introduced, those tools are going to be helping them actually code more efficiently and effectively bending the curve of productivity for these developers. And that's good news for them and good news for the projects that they're working on because that allows them to do projects more efficiently and, in theory, do more projects. But as they're doing those projects, it doesn't change our world because they're actually still managing all of those projects through the same pane of glass, which is Jira and Confluence. And if anything, they're actually doing more projects. So instead of seats potentially going down, for us, what we see is developer productivity increasing for projects that we're working on and more utilization of our license -- of our solutions and not an impact that may be what you were alluding to in terms of a decline of seats. So for us, we actually see a tailwind here. And maybe I'll let Martin comment on -- maybe you want to talk about some of the startups?
- 中性 Rovo 有面向非 Atlassian 用户的 SKU（转录为 skew），管理层更希望这些用户进入 TWC；这是一条不依赖 seat 的 AI 收入路径
We're seeing that. We're putting it at the core of the platform. So that's one really important vector to make sure to understand. Secondly, on the Rovo specifically, we do have skew for non-Atlassian users to get access to Rovo in terms of search and chat and other things. But we think in the medium term, that will be another growth vector possible. We'd rather those users to be on the Teamwork Collection to be actually collaborating, but it's a good stepping stone point to actually get to that. We are increasingly seeing and deploying consumption-based pricing aspects in a whole series of different places.
- 中性 承认存在一派担心 AI 让开发者失业、seats 受影响的观点
The belief around what does AI mean for our platforms because there's one school of thought which worries a lot about, oh, will AI put developers out of work and what happens to seats, et cetera. And then there's another school of thought around what is actually happening on the ground and how is work evolving and what role can companies play in that. At Atlassian, fundamentally, we think about it as AI is going to cause more software to be created. And more software being created is a good thing for us as a business because we are in the business of serving teams that build software. Now the way that teams build software is changing rapidly, who gets to create software is changing rapidly. So it's no longer the technical developer, but also the nontechnical consumer that can create software. So for Atlassian, that's a good thing in terms of how do we serve these nontechnical customers in addition to the traditional technical developer that creates software. The good thing for us is that well over half our business has been serving nontechnical users over the last couple of decades. And so for us we think about the opportunity is really in terms of how do we build these brand-new workflows that serve these nontechnical audiences, which is what we do with Rovo Studio, with Rovo Agents. And how do you really unlock knowledge from various different silos and bring it in service of these new people who are creating a lot of software. In terms of the risk that creates, there is a risk that -- given all the fraud in the market, there is a risk of fragmentation where enterprises end up trying various different tools. But we see it mostly as a cycle. So as people try out new different tools and see what's in it for them and how to use it, typically, enterprises come back to consolidating on a platform and the fact that we can serve different kinds of teams, customer support teams, developer teams, marketing teams, HR teams, we are able to provide that one consolidated platform at Atlassian. So that's the reason we keep focusing on what is that plane of orchestration we can deliver for enterprises to consolidate on one connected data platform.
- 口径不齐 AI MAU 约 230 万（AI 用户，不是付费 seat）
And it's very inspiring to see how values led and how very much the OG culture of thinking for yourself, first principles thinking we've been able to retain as a company. And if I have to think back about what are the moments of inflection points, I would say, one, just our ability to have introduced newer businesses. I remember, I was hired to platformize Jira to go from one product to multiple product and diversify revenue before we went public. And we've consistently introduced new businesses, JSM, Jira Service Management, as a case in point, grown to well over $0.5 billion, 0 to 1 product; JPD, Jira Product Discovery, which has 20,000 customers; and now Rovo, which we introduced Rovo and Atlassian Intelligence a couple of years ago, and now we're at about 2.3 million AI [ mau ]. And very gratifying to see the consistent track record of innovation. And second, the OGs of product-led growth. So like Brian mentioned, we're very much product-led growth, but over my tenure, I've had the opportunity to really expand to enterprise and tell the message of Atlassian's platform for large enterprises and how it can help with their mission-critical work and all of which culminates in the final thing of what we call the system of work.
- 反对 有碎片化风险，企业会去试各种不同工具
The belief around what does AI mean for our platforms because there's one school of thought which worries a lot about, oh, will AI put developers out of work and what happens to seats, et cetera. And then there's another school of thought around what is actually happening on the ground and how is work evolving and what role can companies play in that. At Atlassian, fundamentally, we think about it as AI is going to cause more software to be created. And more software being created is a good thing for us as a business because we are in the business of serving teams that build software. Now the way that teams build software is changing rapidly, who gets to create software is changing rapidly. So it's no longer the technical developer, but also the nontechnical consumer that can create software. So for Atlassian, that's a good thing in terms of how do we serve these nontechnical customers in addition to the traditional technical developer that creates software. The good thing for us is that well over half our business has been serving nontechnical users over the last couple of decades. And so for us we think about the opportunity is really in terms of how do we build these brand-new workflows that serve these nontechnical audiences, which is what we do with Rovo Studio, with Rovo Agents. And how do you really unlock knowledge from various different silos and bring it in service of these new people who are creating a lot of software. In terms of the risk that creates, there is a risk that -- given all the fraud in the market, there is a risk of fragmentation where enterprises end up trying various different tools. But we see it mostly as a cycle. So as people try out new different tools and see what's in it for them and how to use it, typically, enterprises come back to consolidating on a platform and the fact that we can serve different kinds of teams, customer support teams, developer teams, marketing teams, HR teams, we are able to provide that one consolidated platform at Atlassian. So that's the reason we keep focusing on what is that plane of orchestration we can deliver for enterprises to consolidate on one connected data platform.
- 中性 Jira 可以把工作项分配给 agent（GitHub、Cursor、Canva、Box 等），再拉回给用户协作；没说 agent 是否计费或算 seat
Thanks, Ryan. Great question. Firstly, I would say we are making amazing progress in our AI capabilities and delivery to customers. I think the company broadly should be incredibly proud of shipping AI to our customers. It turns up as one of the reasons that they are migrating to the cloud when you talk to them. It shows up as one of the reasons that they're moving to the Teamwork collection. It shows up as one of the reasons they are deepening their relationship as a strategic partner with Atlassian is our core investment in AI, which is world-class. We are doing an incredibly good job at delivering AI to customers to give them the benefits today and giving the confidence that we will continue to deliver that into the future. Jira has always been an incredibly integrated tool. We're very proud of that. The Atlassian platform is incredibly integrated. You can see that there are a number of partnerships that we've signed up in the last quarter, a lot of which are listed in the letter to integrate our platform with other offerings that customers have. It's a core part of our customer value proposition is we believe that your best technology world, your best software world, is a deeply integrated and deeply connected world. You can see that in the ability to use Atlassian's AI off Atlassian from within other tools. and vice versa to use other tools from within the Atlassian world. And I think that will depend on where the customer workflows are and where they are working. Most prominently, recently, you can see that in Jira's ability to assign work items to agents. Whether those are technical agents from GitHub or Cursor or whether those are nontechnical agents from Canva or Box or someone else. This is about taking your Jira workflows and business processes, assigning parts of them off to AI or agents as that world evolves and then bringing it back in for collaboration with further users. So I think we will continue to be integrated. I think that's the best outcome for our customers. And I think that our -- everything from the Teamwork graph to the design expertise that we're putting into our AI offerings to just a world-class set of capabilities that come with Rovo is showing up in our customers, and it is a reason that they are more deeply partnering with us. So incredibly proud of the delivery we've had there and a lot of work to do every single quarter as we go forward. We're rolling.
- 中性 消费制定价是一种选项，客户感兴趣但也谨慎
Thanks Fatima. Look, we have a series of different consumption-based pricing offerings as we have announced and shown from the Rovo and AI credit world to the service collection world of agents and assets to Bitbucket pipelines to Rovo Dev to Forge, CDP, that is certainly something that we have as an option set for our customers. It's one of our elements of monetization. I think it is certainly, something customers are interested in. It's certainly something that they're also cautious about. I think the most important thing for us when it comes to your question about AI monetization, I would say we are already seeing it, our 3 strategic priorities, right? We reiterate them because they are so important to us, delivering a world-class AI platform, continuing to grow our enterprise capabilities and the system of work across our customers' teams and enterprise. We're making amazing progress in all 3, and I will say that it's directly driving the results we see in Q1.
- 中性 市场空头论：AI 使软件从 seat 转向结果定价、导致 seat 压缩
ServiceNow is cheap again. From the current price an investor can achieve a 5-year IRR of 21-29% assuming 25-30x exit multiples, or 16-25% IRR at 20-25x. The stock trades at 24.9x TTM FCF per share (26.7x if we do not take into account the net cash that was likely used for the Armis acquisition).  I previously wrote up ServiceNow [here in June 2022](https://www.valueinvestorsclub.com/idea/SERVICENOW_INC/2785259142). Kindly refer to that post for background. **From the investment in June 2022 through my sale recommendation in late 2024 the stock delivered an IRR of 39.2%.** As I noted when I closed the position, I loved the business but not the price. Since my sale recommendation the stock is down 38% while free cash flow per share has grown 35%. (Note: the stock split 5:1 on Dec 18, 2025.) The stock is down 24% _this month alone_ on the sector rotation from software to semis:  **The macro reason for the stock price decline** is the bear thesis that AI is going to automate tasks that SaaS companies do for pennies on the dollar; that AI will cause seat compression because software will move from seats to outcomes; and generally that new AI-native incumbents will “disrupt” or supplant SaaS incumbents. I rebutted these narratives in my recent [Atlassian write-up here](https://www.valueinvestorsclub.com/idea/ATLASSIAN_CORP/6342411484). That Atlassian write-up also has useful links to help an investor think through these issues.
来源背景：VIC 作者 jgalt（买方）及其引用的 X 帖 - 支持 CEO：协作类软件里客户偏好可预期的 seat 定价，协作人数是目前最好的价值代理
We've done that in storage. We've done that in network costs. We're now doing it in AI costs. And the customers' preferred method of payment is still an understandable, predictable pricing pattern, which tends to be a seat-based model in our category of software in terms of delivering collaboration tools for teams of people, how many people are collaborating is a best proxy at the moment for value, and I believe continues to be so. You could worry about our AI costs. Again, you can see in our gross margin improvements for, I think, probably the third or fourth quarter, don't quote me on that, continuing to improve our gross margin that we're able to deliver those 5 million Rovo seats and continue to improve gross margin. That's a huge achievement on behalf of our engineering teams, but it shows that we can manage those AI costs inside for the vast majority of customers.
- 中性 CEO 认为 Anthropic Cowork 等 agent 工具会加速流程，但不会取消人的参与；Cowork 是 Atlassian MCP server 的大用户
At the same time, you're still going to need human beings in the process in lots of different places. Approval workflows, business processes, they can be accelerated in lots of spots. That's exactly what we see customers doing with Rovo agents and with all of the other agents that can now operate inside of Jira. Those agents are accelerating business processes in lots of different ways. They're not eliminating the human impact, right? The human AI collaboration is incredibly important, and I think it continues to be so. Lastly, I would just say that that's not new for us. Our philosophy of integration goes to listening to customers.
- 口径不齐 agent 怎么变现还没定，是行业内都在讨论的问题
Sure. So as we a couple of things. Firstly, like Jira within customers has consistently been an orchestration layer within customers. And as we move forward in the world of AI, it's going to continue to be an orchestration layer. If that's with humans and if that's with customers using agents, firmly, I can continue to be that orchestration layer. And now as we move forward in terms of the monetization, we obviously have the opportunity to monetize in terms of agents as well. That's something, obviously, that we're looking into how we're going to monetize in terms of agents. That's a debate that we and others are having. And we will see how we're going to move forward in terms of how we're monetizing that in the future and that I know a debate that others are having as well.
- 反对 承认有组织可以自己 vibe code 部分外围功能
Yes. I think like sometimes the best thing I can do rather than talking like in generic about us, like the generic headlines, is actually talk to you like where the rubber hits the road with real customers. So in the past 48 hours, I told you I talked to a technology-based West Coast company, who's clearly entering into a long-term agreement with us. I told to -- I talked to a German automotive customer that has 400,000 seats. Clearly, that is extremely complex and heavily customized, and we will work with that customer to migrate them to the cloud, but they have managed Jira in a way that is supporting their business very, very uniquely. I also spoke this morning to a defense company that told me that we are in every nook and cranny of their business, and we are a mission-critical application to them. When they had a system down, I was speaking to them because they had a system down scenario for their ERP instance, not us, their ERP instance. And when they have that down, they call their GSI, they call their ERP company, and they called us. And they called us because we are mission-critical. We have seen as we've moved more and more companies to the cloud that we're engaging in more conversations with them around topics like SLA. You're engaging in topics around SLA because you are mission-critical. That's a very good position to be in even though these customers are then testing us and pushing the limits in terms of what our service level response time are going to be, et cetera. But this is all because we are very critical to these organizations. Now when you look at customers like the ones that I just explained to you there, when we are so mission-critical, they -- could they do certain things in the periphery to Atlassian? Absolutely. Are there organizations that could vibe code certain pieces? I'm sure there are. However, when we look at scalability, reliability, orchestration management, these are the things that Atlassian brings to the table and has been doing for 23 years. And that's why so many customers, like the ones I highlighted, place their trust in us. And we're very confident that they are going to continue to do so as we move forward, hopefully for another 23 years.
- 中性 DC 客户数量少但 seat 数大，很多有 5 万用户以上
So I think a couple of things to keep in mind is, again, the data center cohort is a much smaller pool of customers, but with much larger seat counts behind them. So that's the first thing that I would point out. I talked about it being a multiyear journey, I had to move these customers over to the cloud. You get varying levels of price uplift to that you're pointing at, depending on how many users the customer has. So on the low end of town, it could just be a cost neutral or 5% uplift. On the high end of town, there could be upwards of 2x the price. So it really depends on kind of the number of users that a customer has. We've been hesitant to give kind of a direct ratio to apply because of that varying level of complexity. But I think that's something that we're thinking more about how do we help investors understand the level of uplift. I think the commonality between all these though, is that because these are the largest customers, we do have a more traditional enterprise negotiation process with them. We want to ease them over to the cloud. So in those instances where they have more significant uplift. We want to smooth that path out for them. So in year 1, it may be less prohibitive and, therefore, make it more cost neutral for them, so more of a 1:1 kind of ratio in year 1. But in years 2 and 3, you start to peel back that level of discounting. So you slowly ramp them to that full list price over time. So you kind of actualize that price benefit in years 2 and 3.
- 口径不齐 MCP/第三方访问怎么变现还在观察
We continue to believe that the open ecosystem approach is an important one. If customers want to tap into Atlassian products, like that's going to be an important one because they should have that freedom to do so. We want to encourage that. What's important, when we look at the MCP data, customers are writing and editing within our products. That's an important aspect rather than just purely reading because that means that we're driving more and more workflows through the platform, and that begets more and more users to collaborate on those workflows and then that also drives upsell opportunity or upgrade potential. So we encourage that type of access into our platform. It is an area that we continue to monitor. I think we look at how the industry continues to evolve about how to maybe monetize this kind of access platform. But again, that open philosophy is a really important one for us. And philosophically, that's something that we believe in.
- 中性 Jira 可以把工作分配给第一方 Rovo agent 和第三方 agent，没提 agent 怎么收费
So the question was, do we see the need for different -- or new products into the mix evolve into this Agentic world. I think we'll look hard at how our customers are using products. I think like actually one exciting development with Jira that we just announced, I think, 2 weeks ago now is the ability to assign work to agents directly in Jira. That includes not only our first-party Rovo agents, but third-party agents as well. So for example, we're working on something, and I say, let's spec out and drop the product specs. You can assign that to a sigma agent today. You can sign that to a Rovo agent today. So that, I think, is an exciting step back to that point that we talked about earlier of you still need to track and manage where all this work is happening. And even if third-party agents are doing some of that work, you still need a place to track all that complexity, especially with -- as this continues to grow. And so I think that's an exciting aspect of Jira placed as kind of this pane of glass of tracking where all those agents are doing and what are they actually working on? And are they actually delivering the results that we wanted.
- 反对 Atlassian 2-3 年内必须从 per-seat 转向用量计费；已有 assets、automations、Rovo 三部分按用量收费
SD: All right. So, big picture. I was going to ask you to define that. But a couple of years out, I definitely think that we're going to see a much evolved Atlassian where they're making their revenue differently than they currently are. So mainly they're a per-seat revenue company. I don't see that staying the same in 2 years or 3 years. I see that they're going to have to change it. The consumption model is pretty much what most SaaS companies are moving towards. We're seeing some of it from Atlassian. There's like 3 aspects of the Atlassian suite that is now consumption-based, which is pretty much assets, automations, and Rovo. So the reason for that is… AW: Shawn, do you think that their migration over the next couple of years to a consumption model will be more profound or more impactful generally than everyone, right? I mean, it seems like with the advent of AI and AI native solutions that whether you're Salesforce, ServiceNow or Atlassian or other, that we're finally seeing acceleration away
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 反对 剔除强制打包后 JSM 底层需求没变；小客户 JSM 放缓，部分因为用 AI 自己写服务台更容易
SD: Yeah, it's kind of the same. Where we're seeing the wins on JSM is customers that aren't using anything for service management. They're definitely small to medium-sized businesses that we don't necessarily target. It has slowed down a little bit in Q1, but that's probably chalked up to 2 things. 1, it's easier to AI code a service management solution for the small mom and pop shops that were kicking the tires of JSM. And 2, the marketing has slowed down. The marketing message from Atlassian as well as from partners isn't pushing, you know, we're not pushing JSM as hard as we were the last 2 years. After Teams, depending on, you know, the Teams, the big conference coming up in a couple weeks, depending on what's announced, unless something earth-shattering is announced, we plan on shifting back to pushing our JSM offering after Teams. We're kind of waiting to see. Right now, we're just promoting Teams to see if we can help drive attendance to the conference. And so that's our main message as well as we just did 100 videos about concerns around moving to the cloud and how we would solve that. So we're really pushing the cloud migrations and the conference in our marketing. And so that's our reasoning or how we justify a slight slowdown in our JSM services.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 中性 GitLab 非 PubSec 没有看到 seat 减少，但也没有大幅增长
SD: Yeah, we didn't see any reduction in seats. We didn't have a lot of expected renewals in Q1, but the ones that we did came in on time and pretty much where we expected it to. So I can't say that we're seeing a huge increase in GitLab for the non-PubSec, but we're not seeing people running away from the product.
来源背景：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防。Q1 的 seat 激增集中在一家国防客户 (300 → 5,000)，他本人怀疑对方会不会真招满 - 中性 Jira 仍是 IT 支持团队主工单系统，Console 位于 Slack 与 Jira 之间
There's a couple of stages to that. We still do have Jira as our main IT support help desk for the IT support team. Console sits in between Slack and our help desk. If somebody submits it via email or directly through our portal, it just goes directly into Jira that doesn't get touched by Console. We do have some automations that we can use to basically redirect them to Console to use Console first, if it's a request or a ticket that we know Console can handle. There are plans to potentially move away from that model and move directly into Console as our one-stop shop. Not only for IT, we actually have plans potentially to introduce this particular tool to other support teams within our company. A lot of our team members have internal customers and Console really excels at that, answering questions, directing them to the right place and then also submitting tickets on behalf of users. This year we're actually looking to expand that into other teams that can include like finance, procurement, onboarding and offboarding HR, even our security team, things like that. That way, somebody goes to a single source and says, "Hey, I have this question or this request or this whatever," and they'll be directed to the proper authority for that particular request based on the build out that we do for each of those different teams.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户 - 反对 Console 自有 help desk 可为不需要 Jira 复杂功能的团队完全替代 JSM
I think they're fairly large in terms of potential, because they can actually replace several different tools. They can replace Jira completely. Console released their own help desk, which works really well, especially for teams that don't really need the robust nature of Jira. You don't need a bunch of other automations, you can just use Console and it can replace Jira Service Management for a number of other teams. It will also improve SLAs. It will allow for better tracking of data and documentation. It will also track when we're missing pieces of information that should be built out. For example, if a ticket continuously gets escalated, it can be flagged and say, "Cool, we need to build out a workflow or SOP," or things like that. Price-wise, I think it'll probably be significant once it's fully built out and it does everything that they want it to.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户 - 中性 Console 按全员计费约 $10-15/人/月，年费 $100-200k；可接受上限约 $20-30/人/月
When it comes to the pilot, basically it's everybody can access it if they want and they actually recommended that "Hey, just let everybody do it." We were concerned, "Hey. What if this product sucks and we release it to everybody and all of a sudden we have to say, 'Oh sorry, just kidding.'" What we did, we did a 500-person pilot and that just means that we notified 500 people that they could use it. Other people did find it in the channel that we were using or were invited by other people that were in the demo. We quickly realized that there was a good product. The contract itself is for full-time employees. Basically, it's for full-time employees that are in our environment. We basically look at our Okta instance or our [Google](https://app.tegus.co/app/company/51/alphabet-inc) instance and that's we pay per user. I think at this point it's somewhere around the $10- $15 per user per month to the amount that we have in terms of team members. I think at this point it's probably between $100,000 and $200,000. That's going to depend on a couple of things. They have different price structures based on if you just have a single IT help desk or us we actually want to expand it, so the new contract will be a higher rate because we get unlimited workspaces. We can make a different workspace for each team that needs it. Basically, there's a set fee to go to that unlimited status. I don't know if that's going to be forever, but at least in this contract it is.
来源背景：Senior IT Systems Engineer at Flock Safety，Console 客户 - 中性 Jira Align 定价在 Atlassian 企业协议层面打包，但仍间接随 seat 数扩展
EXPERT: In enterprise PPM tools, pricing is almost never truly transparent, and it's typically structured as a subscription-based enterprise license, but with different metering models depending on the platform. In our evaluation, Planview and Jira Align were both priced as enterprise subscriptions, not simple per-user SaaS, typically negotiated as an annual contract based on a combination of total users, modules enabled, and overall portfolio scale, effectively making them platform licensing deals rather than strict per-seat pricing. Jira Align still indirectly scales with seat counts, like portfolio users, contributors, and read-only users. But pricing is bundled at the enterprise level with Atlassian agreements. Planview followed a similar enterprise subscription model but was even more heavily tied to modules like financial planning, capacity planning, etc., and organizational size, making it the most rigid and highest cost structure. Smartsheet, in contrast, was more clearly per user, like per member, per seat subscription model with additional paid add-ons for advanced portfolio capabilities like Control Center and DataMesh, making it more modular and scalable in cost, but less predictable at enterprise scale.
来源背景：Mahak Varshney，Canadian Tire Product Manager - 中性 CEO：目前最大部分的价值通过 seat 定价交付；另有 10-12 个消费计量项
Our philosophy has always been to meet customers where they're at. Let me start there. Customers, in general, like the way we price our offerings and we wish to continue to be customer-led and meeting them where they are. The largest amount of value delivered today is through the seat-based pricing model. The collections have been a major transformation in how we do that, for sure. And in that, you are getting a broader amount of value that you can see there.. So we talk about that when people move to the Teamwork Collection, which we're seeing great momentum in.
- 中性 第三方 agent（Gamma、Canva、Cursor、Claude 等）可在 Jira/Confluence 中运行，都调用 Teamwork Graph；没提计费
There is no doubt that agents existing in native context in our automation framework. If you have a huge amount of business processes running through Jira or the Service Collection, the agents that are natively integrated have the largest access to that platform and they're right in the sidebar. They appear in the Jira UI, that's a huge advantage. At the same time, we've shipped a whole series of features that allow third-party agents from Gamma and Canva to Cursor and Claude Pro in each of our different types of teams. We want to make sure that third- party agents also surface in Atlassian's context, whether that's on a Confluence whiteboard, whether that's on a Jira work item, but they all use the Teamwork Graph at the core. So customers are really seeing that. We certainly get evaluated against other platforms.
- 中性 管理层主动点名市场担忧：AI 让开发者多余、seat 增长会死、seats 会消失、货币化模式不对
I think it is totally true. There is a big disconnect, I believe, in how we are seeing and what we are actually delivering. And so I thought we'd go through some of those doubts and give you a lot of facts and figures and other things to go through those. So you've probably familiarly heard a bunch of the AI is going to make devs redundant. We only serve devs. Seat growth is going to die, seats are going to disappear. AI agents are going to make everything less critical. People are going to, I don't know, make their own SaaS by themselves, make their own operating systems, don't probably make everything. AI native start-ups are going to just kill everybody. We don't have the right monetization model.
- 中性 大多数 app 仍按 seat 定价，CEO 认为这是多数客户想要的购买方式；同时存在用量、混合（TWC）与承诺制（flex）模式
We have multiple pricing models. We're trying to be customer-led. We're trying to meet customers where they are. Most of our apps are priced per seat. I still think that's the way most customers want to purchase software. When I talk to customers, that is what they say they want. We have a very generous number of Rovo Credits involved in those per seat accounts. So we have the price per seat. I'm sure you're familiar with that part of our model. And these are sort of illustrative down the bottom.
- 口径不齐 正在与大客户谈 flex 承诺制模式，允许在 seats、apps、collections 之间挪用承诺金额——未来 seat 数可能不再是收入的直接口径
So it's a different way of thinking about it. And we have a flex-based model that we are working on with customers at the moment that we're actually talking to a whole bunch of customers this week, which is more of the overall sort of commitment-based model, as I would say, is the way to think about it. So as our bigger customers that get well north of $1 million, say, here's what I want to commit to your platform. I want to be able to move it around seats, move it around applications and collections and I want more usage-based certainty. That is another model that we have. Again, we're here for the long term. This is us being customer-led as they ask.
- 中性 大客户案例：汽车制造商从 5,000-10,000 用户增至云上 70,000 用户，ARR 从 1-2M 增至约 9M
And then you see as they move to cloud, they continue to ramp up. We talked a lot about why that is, but it's fundamentally, they have many more things they can buy. Their expansion opportunities, their expansion speed in the cloud is a lot faster, and they've moved to many more users. Back here, they're in the 5,000, 10,000 user range. At the end now, they have 70,000 users running on cloud enterprise. And the scale of what they're doing is absolutely wild, right? Millions and millions, in fact, more than 10 million automation rules running every quarter, hundreds of thousands of manual tasks disappearing, massive Rovo adoption across the organization, agents running AI. Rovo is their default for AI.
- 中性 DC 仍有大量 seats 尚未迁移，迁移仍有空间
I hope you can see about our multiyear decade-long cloud transition, both in terms of the building of the future ahead and moving customers. I think we've executed extremely well across the hundreds of thousands of server customers. Now we're in the data center customers at some scale. But the migration itself still has a ways to go in terms of the number of seats we have in the data center. That's a good sign. What does this look like? What is an illustrative example of customer pricing? It's always hard, so I picked an example. 93% of our data center customers that are upgrading to the cloud are landing in premium or enterprise. And I know this because in some of the models and some of the things we've seen, you can think, oh, well, the data center customer could go backwards moving to cloud standard. That doesn't happen for 93% of those data center customers can mathematically happen, but I thought it was useful to say this.
- 中性 Teamwork Graph 不单独收费，作为平台一部分赠送
I think, look, firstly, customers don't have to buy the Teamwork Graph. They get the Teamwork Graph. We've made it a core part of the platform. There's a reason that we don't sort of separately charge for it or sell it or any of this sort of stuff, which always makes it harder from an investor point of view to communicate that. You'd love to see a separate line item that's going up into the right and it's looking really, really great. But from a long-term strategic point of view, I don't think that's the right thing for us to do. What we want is anybody using Jira, anybody using Confluence, anybody using Loom, anybody using Bitbucket to have the Teamwork graph.
- 反对 大客户裁员导致 license 下降、产生基于金额的流失，压低 GRR
**Expert:** SmartBear's largest customers who were going through just workforce reductions, there's definitely dollar-based churn when licences were going down. There was certainly sensitivity on GRR level if customer lays off 1,000 people and 100 are developers. You just don't have as many seat licences to sell. That was not the majority of customers that was impacting several of large enterprises and a bit of downward pressure on gross revenue retention as they just don't renew with as many licences. **LF:** Is SmartBear's revenue becoming less sensitive to who is doing the work, and more sensitive to the volume of work? **Expert:** Yes, and I think that then becomes the pivot of moving away from just that seat licence is now the volume of work is going back up again, even the number of developer or just people with the new titles, but they're focused on delivering software is increasing and shifting that churn sensitivity goes down. It becomes more stable.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 反对 客户缩减表现为 seat 数减少，而非降到低档
**Expert:** I think it's going to trend through existing cross-sell and expansion and adoption of BearQ and AI. New logo growth, again, it's somewhat small dollars for the net-new because most of the large enterprises are using some extent of the SmartBear tool. It's hard to count a big enterprise is a new logo. They tend to be the smaller ones or initial small lands. I think it is going to be around the cross-sell motion in terms of like… If you're asking my prediction of going from USD 240m to what, I think there'll be fairly low top-line growth for a while until the promise of the BearQ starts to really accelerate in the market and the market sees the quality requirements coming out of AI-developed software and apps going up. I think there'll be a bit of like a U curve, not a downward U, but just a curve of revenue growth. It might be a little bit on the low side for a while until the transition is done is my best prediction. **LF:** You would estimate something along the lines of mid-single-digit revenue growth. Is that fair to say over the next two years before revenue from BearQ is realised? **Expert:** Yes. Over the next year. Hopefully, it's not two years, that starts accelerating again. **LF:** Are you seeing an increase in downselling, where customers move from premium versions to lower-tier versions during budget reviews? Is that something that you see or expect to see? **Expert:** I didn't see that. It was more just the number of seats went down. I didn't see dropdowns to lower tiers at the time. I can't speak to current.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 反对 内部讨论认为 SaaS seat license 定价正在消亡，需转向规模/用量混合定价
**Expert:** When I was still there, we were having a lot of discussions about completely changing pricing models because when you see on the horizon that the seat licence pricing model for SaaS is a dying thing, and using a pricing model that worked in 2021, nobody is there anymore. I think just shifting to some level of scale-based, limit-based, usage-based various hybrid price models is going to be their play to quickly pivot to the way like newer software is being marketed and sold is important. **LF:** Do you think SmartBear will ultimately be successful with that pivot? **Expert:** They have to be. The folks there are smart enough to recognize it, so they're going to have to make that pivot. Whether they completely just rip all the Band-Aid off and reprice everything immediately or not will determine how well and how quickly it gets executed and transitioned. I think that's a bit of a paradox of they may take a short-term hit for a long-term gain by completely just ripping and replacing right away or slowly transition. Again, I don't know which one is the best or which one they'll go with when they do make that transition, but it does have to happen.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 中性 Zephyr 等与 Jira/Atlassian 生态深度连接，黏性高
**Expert:** I think the stickiest products include the test management suite, Zephyr, QMetry as well as TestComplete, which becomes highly embedded in workflows, SwaggerHub. Swagger has been historically very, very sticky, but also it can be lower level like single departments that stick with it. Getting back to the test management side, test cases and history and just release automation gets very, very embedded in the toolset. Once an organisation has really anchored on their test management process, it's a lot of effort and time to switch to something else, and those products just tend to work in the Zephyr suite and the QMetry suite. There's no reason to make a switch there. They also have been connected to Jira and the whole Atlassian ecosystem. Those interconnected connectedness of that plus all the repositories make it very, very sticky. TestComplete, same thing. TestComplete and Reflect, those testing automation tools tend to sit within the SDLC workflow and with the automations built, it's difficult to rip and go to something else.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 中性 AI 不确定性下企业把软件合同期缩短至 1-2 年
**Expert:** I think there will actually be a short-term duration where contract lengths are shorter as the new world of what software enterprises invest in, and AI shakes out. I'm seeing that across the board with investment length being averaging one year, 1-2 years instead of 2-3 years. I don't think that will be long lasting. I think it will come back to longer-term contracts to lock in some price protection, but just next year or so, we'll probably see a compression of contract length. I think that's true for SmartBear as well as every other software company, including AI tools. As the price of AI just using the tokens and Replit or Claude or whatever it is, folks are using, starts to increase, people will come back to locking in longer-term contracts. **LF:** Is this change in the contract length driven by customers rather than companies such as SmartBears? Is that where the demand for this contract length shortening is coming from? **Expert:** I think it's driven by customers as they're testing and trying new things and new ways to work and not sure it's going to stick or last.
来源背景：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初） - 支持 Jira 在工程端根深蒂固，竞品难以渗透；价格 $7/用户/月且覆盖 80% 功能
example, very quickly finance would realize how much more beneficial it would be to get their colleagues on the tool. And so it would sort of spread virally from 1 or 2 departments. The main functions where Asana did well were the functions where a lot of the work was templatized. So for instance, marketing. Or product, and what I mean by templatized is these functions would go through the same sequence of steps to do similar work. So, every time there is a new marketing campaign, you need to design the creative, get legal approval, get finance, get budget, and pass the leads on to sales. It was... I wouldn't say cookie cutter, but there was definitely a sequence of steps that was relatively consistent across pieces of work that would make Asana far more beneficial for them. As opposed to a function where no 2 situations are really that similar, they wouldn't find as much value in the templates that Asana would provide. But generally, it would start with some of those more sequenced functions. The 1 where we didn't really do well was engineering. Engineers often operate in their own... like a walled garden, to use a term that people use to describe companies like Apple, for instance. They've got their own tools, they operate in their own cadence, and Jira from Atlassian was so embedded in the engineering community that it was very difficult to penetrate. Jira is quite clearly made for developers and engineers... they advertise it on their own website themselves... but it's very difficult to sell Asana for $33 a month when Jira is $7 a month per user and does 80% of the functionality. So yeah, that was generally speaking some of the functions where we did well and where we would expand.
来源背景：Robbie Goldberg，前 Asana APAC Corporate Sales 负责人 - 中性 企业端微软/谷歌/Atlassian 的捆绑让独立工作管理工具难卖
RG: Yes. Asana is definitely doing better in the enterprise space and I could just tell that based on a lot of the hires they've made. Asana is hiring a lot of people from Salesforce and like the Microsofts of the world, the people who are used to selling to these large enterprises, so they are trying to establish credibility through its personnel because a lot of people when they are making a decision about this kind of software, they're not just looking at the functionality of the software, they're looking at the partnership with the particular vendor, they're looking at who can help them deploy the software and manage the implementation. So the people component is very important. But one of the limitations has been around... it's just so difficult to compete with some of the larger companies who can bundle Asana's functionality for free, so like a Google or a Microsoft for instance. When you are buying Google or Microsoft, they can make it appear that their collaboration functionality is free, like Microsoft Teams is free; no one pays for that, they essentially pay for it because they are paying for their office suite, so Word, PowerPoint, Excel. They're paying for Microsoft Outlook. And then Microsoft tells them that they just give them Teams for free. So if you're looking at a budget line item and you're the Asana sales rep and you're costing $33 a month all the way up to $70 a month and the Microsoft equivalent looks free, it is a very, very difficult sell. So one of the biggest threats is just commercial. From an actual technical perspective, Asana has single sign-on, it's got data centers all over the world, so data sovereignty is not really an objection. There's all kind of enterprise-grade security and ISO 27001 certifications, all that good stuff, but commercially it's very difficult to sell to the enterprise against the likes of Microsoft and Google and Atlassian.
来源背景：Robbie Goldberg，前 Asana APAC Corporate Sales 负责人 - 反对 UBS 调研中 Atlassian 是最常被降优先级的类别之一，seat 优化与价格抵抗成为新逆风
UBS says recent enterprise software checks point to worsening IT budget conditions for SaaS vendors, with customers increasingly prioritizing AI, cloud infrastructure, cybersecurity, and data spending over traditional seat-based application software. The firm highlights Workday, Salesforce, Atlassian, and HubSpot as among the most frequently deprioritized categories, while shorter contract durations, pricing pushback, and seat optimization are emerging as new headwinds to cRPO/bookings growth. UBS continues to favor AI/data infrastructure beneficiaries like PLTR and SNOW, arguing “build vs buy” dynamics and internal AI application development are creating additional structural pressure on broader SaaS spending.
来源背景：UBS 卖方（TMTB 转述） - 中性 扩张从'买更多 seat'转为'seat + 用量'，可在已有部门内通过用量加深
JS: Sure. So usually expansion within Monday is led within an organization when other departments sign on to use monday.com for their workflows as well, and it's a very natural progression inside of a sales cycle or even for a customer if every department in a business touches another department. There is always going to be a point where that handshake between departments is recognized that things would improve with a software that everyone was on the same page with. So before though, that used to mean a customer needed to buy more seats, and this was the name of the game in the SaaS space for the last 10 years. Very true at Monday. But now, that means more seats and usage with the rise of AI and the ability to sell usage. So this is a very interesting way to monetize, and it's going to be very interesting because expansion might not need to come across departments. It could actually begin to grow within a department that already had seats, but now is deepening their usage. So it's a really nice way for Monday to expand and increase their revenue and go much deeper into the customer base they already have.
来源背景：Jessica Saranich，前 monday.com Regional PMO Director & Chief of Staff - 支持 Monday Dev 比 Jira 弱很多；纯开发客户中 Jira 会赢
JS: Sure. So similar to my answer before around, you know, how competitive, how tall can Monday stand against its competitors? If Monday was competing against only Atlassian, who I believe still owns Jira, I would say they're very vulnerable. If the customer is only a developer shop and that's their main tool, I could see Jira really just taking over and winning the customer there. ServiceNow and Salesforce, not so much. ServiceNow is quite expensive and really stands tall in the IT system use cases. Whereas Monday is a much more broad offering than that, more flexible. Salesforce as well. Monday CRM, I would not say is a direct competitor to Salesforce, but does stand tall in its own right. And really with the stickiness that Monday CRM also enables the customer to work beautifully in the Monday Work Management platform as well. Microsoft is a big risk though due to what I mentioned before being quite expensive and a legacy tool that most companies already have.
来源背景：Jessica Saranich，前 monday.com Regional PMO Director & Chief of Staff - 中性 Atlassian 的 Teamwork Graph 比 Smartsheet 知识图谱早两年以上
NA: Yeah, so I'm afraid that at this point, from what we know about it so far, it's not a differentiator. It's very much about catching up to competitors because if we simplify what Smart Hub does for a second, so Smart Hub allows customers to create custom agents that are personalized for their particular workflows and their particular ways of working. Asana, for example, already has Asana AI Studio available where those types of agents can be set up. Monday.com has an AI Agent Builder, very similar capabilities, and Atlassian has Rovo Studio. So, yeah, it's very much about catching up to those kinds of capabilities. And if you listen, for example, to the last Engage 2025 customer conference, something that the CEO mentions is that one of the most important components of Smart Hub is the knowledge graph, which is that contextual data layer that will actually feed it so that it has the contextual knowledge to provide relevant outcomes to the users. So I would agree with him on that but what's interesting is that again it's been announced in Engage 2025 when for example similar organizations like Atlassian announced the Teamwork Graph, as opposed to the knowledge graph that Smartsheet has, and the Teamwork Graph was announced in April 2023. So there's more than a 2-year delay on that contextual data layer that is, according to the CEO of Smartsheet, the most important part of the AI capability. So a 2-year delay is quite substantial, so it's mostly about catching up to competitors. Yep.
来源背景：Nordin Abajtit，现任 Atlassian solutions engineer、前 Smartsheet（可能偏向 Atlassian） - 支持 Monday 很难转化长期用 Jira 等企业产品的客户
JF: They're just well-respected, and I think that to try to break into that and to be a real alternative has been more challenging. I think that Monday Dev is a great answer for smaller dev teams, like I said, less kind of process-heavy. And yeah, I think it does have its advantages that it is less complex to set up, it's easier. Again, the time to value would be much quicker in terms of onboarding a team onto Monday Dev, it would be much quicker than onboarding a team onto Jira. So I think that it's... Jira and Azure DevOps, it's more enterprise and Monday is more small to mid-market, and I think some of the others, maybe Linear is the one that we can compare to most. So I think that you gain the Monday suite of products with it sitting with other products, so Monday Work Management or Monday Service, for example, whereby cross-department collaboration can occur within the same system. I think that that is what's going to set Monday apart within that small and mid-market customers or smaller dev teams in companies than Linear or something like that. But I think that, again, it is more challenging for Monday to convert customers who have been working with Jira or some of the more enterprise products until now.
来源背景：Julia Fagelman，前 monday.com Customer Academy Manager - 中性 seat + consumption 混合定价可能成为 AI 收入驱动
JF: I think it's both, to be honest. I think they needed to show their existing customers that they are making the shift to AI in order to retain customers. It's partially retention and obviously showing that they are moving into this AI category. But I think that there is potential for these to become big revenue drivers, especially for existing customers who, if they start to really use agents within their Monday workflows that they've already built out, especially in bigger companies. That's going to be another reason that it will be difficult for them to churn. And I do also see that it can be a revenue driver as well, especially if the pricing switches to a combined seat and consumption model. Especially in companies that are going to adopt these, especially companies that are going to adopt the agents feature, if they play it right, I think there is a real chance of growth there in terms of revenue, especially for existing larger enterprises.
来源背景：Julia Fagelman，前 monday.com Customer Academy Manager - 口径不齐 AI 货币化当前走 TWC ARPU，未来可能更多转向 consumption（与 seats 并行）
I think different companies are on different arcs of that right now. I don't think it's one size fits all. I think what we're hearing writ large from our customers is, as it relates to AI, is they do want to unlock AI. We're seeing that in our performance in things like Teamwork Collection, where they get 10x more Rovo credits and the growth and opportunity that we see there. But they're looking also for predictability, right? We're hearing a lot right now about token maxing and making sure folks are being really mindful about the tokens that they're spending right now. And so they're looking for predictability. And they're also looking for the ability to load balance those AI credits as well, right? So for any enterprise and organization right now, oftentimes, what we see and what they see is that they'll see hyper users of those credits, and then those that are deployed, but not necessarily using those credits. So it's helpful for them to be able to take a package of credits and tokens and being able to load balance that across their organization as well. Right now, for Atlassian, the best vehicle to monetize AI and the best way for our customers to unlock AI is through Teamwork Collection because when they buy into that collection, they are getting 10x the Rovo credits. And we're seeing a lot of traction with that right now. We see that Teamwork Collection customers are using 2x more credits than those on a stand-alone basis. They're deploying 2x more agents, which then in turn consume more credits as well. And for Rovo customers, we're seeing credit usage grow 20% month-over-month. So a lot of traction on the AI side there. It's being monetized, and we're seeing that today through ARPU in our Teamwork Collection. And I think over time, that may shift over more to consumption, along with seats. And we also introduced our Flex program as well. This is a single wallet for customers, our largest really strategic customers to go all in on the Atlassian platform. That's access to the entire suite across our collections, across all of our applications and being able to unlock all of those credits across all of the different applications and load balance against those applications in terms of the value. And so through Flex, they're able to actually drive the type of value that is most meaningful for them across the entire platform.
- 反对 seat 缩减是其在 GitLab 及现职公司亲历的普遍问题，一年后 seat 定价将毫无意义
**Expert:** Seat count reduction is… It was like what I experienced when I was at GitLab as a manager and leading a team, and I saw it across the region, across the US and it continues to be an issue even in my current work, that's not GitLab. Yes, it's pretty apparent, it's 100% certain by now, in a year from now, seat-based pricing makes no sense. Everyone pretty much has to move to consumption. GitLab is already in the process of doing that. They called it out multiple times. They hired a CEO to literally do the same thing. He helped New Relic years ago move from seat-based to consumption-based. He's doing the same at GitLab. It just takes a little longer because GitLab's infrastructure isn't set up to do consumption-based pricing yet, but they're already pretty much like 25% of the way there, if you look at all the stuff that's released.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 反对 NRR 显著下滑，主要受中小客户拖累，企业客户较稳
**Expert:** It's dropped. If you look at the financials, it's dropped significantly YoY, the net retention rate. I expect that to continue dropping. A lot of it is just like the larger organisations, GitLab is still going to keep. You have the federal government, no one is going to churn overnight. Where the net retention rate is getting hit by is actually the mid-market, the smaller companies because they're more prone to one pricing, they're more prone to they have smaller teams, they have less complexity. They have less things to migrate. If it really comes down to it, if they really want to move off and churn, they can. I think you're going to continue to see that because that's what I saw when I was there. The enterprise segment typically was pretty steady. We have one big customer that would churn once a year like FedEx, for example. FedEx churn because they hit at AWS and GitLab was partnered with AWS. It's just one of those things where you're going to see that NRR get hit mostly by still the mid-market companies in general.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 中性 IT 预算总盘子不增，更多投向 AI，非 AI 工具销售周期变长
**Expert:** I don't think it's the bigger deals. I really think that it's literally every company is saying that. I don't think budgets are going up for IT in general. It's the same size of the pie, just putting more towards AI now. There's going to be less slices for companies like GitLab or really any company out there. What that means is instead of an easy renewal or easy land a new customer acquisition deal, you're fighting with your champion to your CFO like, "This is why we need this tool that's not AI." Because CFO is going to always ask the same question like, "I'm being told by the Board to adopt AI. We're putting most of our budget right now towards AI. Tell me why we need this tool and tell me why we need this much of a tool." It's just harder to sell any tool that's not innately AI, and that's why deal cycles are going to be longer.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 中性 专家认为 Atlassian 不是 GitLab 的竞争者，只做项目管理
**Expert:** It's a good question. I know from what I remember, the government, they only sold Ultimate. I think any customer they landed is already on Ultimate, but there was no option other than Ultimate. I don't think there's much upsell opportunity really. I think they already acquired most of the agencies by now. I think that group right now is mainly just… They're not really landing more users, probably just all playing defence right now of like, "Can we keep this contract at the same level?" **LF:** Who do you think are GitLab's biggest competitors within the federal space? **Expert:** I don't think there really wasn't a competitor in the public space. GitLab is almost the de facto standard. Only competition is really with itself. Is that agency and the federal government willing to use a free open source tool that's not supported? **LF:** How do you view the threat from Atlassian? Do you see the company as a serious contender to GitLab? **Expert:** No. Yes, not at all. They never really were because Atlassian just does all their project management stuff and that's it.
来源背景：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt - 反对 知识库评估中 Notion 第一、Confluence 第二，Confluence 被认为老旧、行动慢
**Expert:** The whole process, as I had mentioned, lasted three weeks. We tried to do it quickly because we didn't want this to linger on. We were able to get things up and test multiple systems at the same time. **Client 1:** From a task automation perspective, it is well understood that you preferred Ema because of what it offered out of the box. **Expert:** The next one was going back to the knowledge hub. We evaluated three different companies, Notion, Guru and then we already had Confluence, so we already knew that. The way that each of them was selling themselves was more of as we were talking about before, the company brain or a lot more of documentation. It was more of a one-stop shop. Notion competed much more heavily with Confluent in terms of what Confluence is trying to shift. I mean it's very old school. The data in there is horrific. Their agents because it's so legacy, the agents that they've created, they're moving in the right direction and that it is a big competitive threat because you still got to displace the incumbent, and they're showing their agent capability heavily, everyone is. Now the question is, do you want to wait until it meets your needs? Do you want it to move quickly? The big reason there was we were able to abstract the information, getting the information into notion, having a one-stop shop for us to be able to write agents, set up automated workflows, have it be very flexible. Now in some cases, it was extremely flexible and it was hard to use. One of the things that we loved about Notion was the fact it could be a one-stop shop to help our teams across the board, in particular, the account management team and the sales team loved the way that Notion AI and agents were able to set up and automate a lot of their flows. Notion from a knowledge hub perspective was number one, then Confluence, although they were moving slowly and then Guru because Guru is still green, if you will, from a knowledge hub perspective and a one-stop shop compared to Notion.
来源背景：前 Symphony Talent 高级产品岗位 - 中性 定价模式不会一刀切，seat 与 consumption 都可能，TWC 提供可预测性
Yes. Good question. I think I heard maybe 2 questions in that, first around sort of pricing positioning, pricing power around that. Historically, Atlassian has been very transparent with regards to our pricing. And we've always felt really good about the ROI that we drive for our customers. We consistently hear that from our customer base. And I think that's partly what's driven the ability to take share over time and be able to expand across the enterprises as well. It's one of the #1 things that we hear when it comes to the displacements that we've seen over the last several years here as well. One thing that -- I think, again, that I'm going to really appreciate is the ability for Atlassian to take the feedback from customers in terms of product, product parity, features, et cetera, be able to iterate on that quickly and still have really compelling price points to make sure that we're driving that adoption across enterprises and customers. So feel really good about the position that we have across that. And if you think about, again, the -- 85% of the Fortune 500 are customers of Atlassian, but that's only 10% of our revenue. It gives us a ton of room to continue to grow. The other area, maybe to your question here around like how that evolves over time. As it relates to commercial models, right, whether it is a per seat model, whether it's a consumption-based model, I don't think it's one size fits all. I mentioned earlier that Teamwork Collection is one of the best ways that our customers want to buy. And part of the reason of that is because customers are looking for some stability, right? As I'm sure you all have heard the term token maxing out there right now. And one of the things that we hear from our customers is, listen, we want some predictability in our billing structure, right? So we want to be able to buy in and unlock AI. And so they're doing that through Teamwork Collection because they're getting 10x more credits right now. But over time, as that consumption continues to grow, as those seats also continue to grow as we saw in Q3, right, Teamwork Collection outperformed as well as Jira stand-alone seats outperformed, and that was across both developer and engineering seats as well as the business seats as well, we'll continue to evolve the business models and the commercial models to meet those needs.
- 中性 Jira 中可以把工作直接分配给 agent（未谈 agent 是否收费/算 seat）
Yes. I think certainly, Q3 numbers kind of back up a lot of what we certainly have believed. And I think when you think about the complexity of the world and the speed in which AI evolves things, that -- the need for collaboration, coordination raises more than ever, right? Like you need to coordinate more across what are we doing across the organization, especially as the speed and throughput increases and certainly, not just coordinating between humans, but also coordinating between humans and agents. Are we doing the right things? Are those agents actually delivering what we want them to? If Gregg and I are on the same team, we're making sure that as we deploy agents, we're not overlapping, having kind of conflicting things that we're working on. Are we actually progressing against our strategic results as a company. And so that coordination aspect becomes more important than ever. I think we see more people needing to be added to the Atlassian platform to coordinate that. And then you see things like our innovation of adding or being able to manage and orchestrate agents in Jira. That's a prime example. Jira today is already the orchestration workflow platform for work. Right now, I can assign work to my colleagues and team, and now I can assign work to agents directly in Jira and then the agents can take action directly from there. And again, so managing what are our teams working on, managing now, what are those agents working on? Are they delivering actual tangible results for us and advancing our strategic priorities as a company because that's actually what matters, right, and not just the volume of work.
- 反对 渠道调研显示 Jira 向 Linear 丢份额（竞品挤占）
Citizens maintained a Market Perform rating after channel checks, with 11 of 18 key data points coming in negative, led by evidence that Jira is losing share to Linear while the wind-down of Atlassian’s Data Center business remains the largest structural headwind. The firm highlighted encouraging momentum in Rovo, where AI usage is growing 20%+ m/m, and believes Collections remains an attractive monetization opportunity, but FY27/FY28 EPS estimates remain below Street. Citizens argues the stock’s 2.6x CY27 EV/revenue valuation appropriately reflects slower growth and competitive pressures.
来源背景：Citizens 卖方（TMTB 转述） - 反对 Bitbucket 在 AI 代码审查等投入少、逐渐萎缩
So, for example, GitHub native: GitHub has been working on CoPilot metrics, so insights, basically, so that you don't have to rely on a third-party tool. CoPilot review, they have their own review that's much more powerful than it was just a couple of months ago, and all platforms kind of do that. Like GitLab has developed more insights as well; GitLab has developed more AI capabilities, review capabilities as well. The only company that hasn't invested in them that much, I think, is Atlassian with their Bitbucket sweep, basically. As far as I am aware, that's kind of like dying on the vine, and I think they're okay with that. So that's the only place where I would say that LinearB innovation is faster than Atlassian's innovation. Let's see, have I missed? And of course, the competitive landscape is still like tools that are specifically focused on code review, like we already discussed, CodeRabbit, but there's a bunch of other ones in the market, like smaller ones that are laser focused on doing just that. So by itself, if you're looking at LinearB's code review unbundled, there is not really any competitive reason to sell. They wouldn't just sell code review, right? They could only make sense as a capacity in a box full of other things that might have some value. I haven't explicitly mentioned workflow automation, but my opinion is that all the workflow automation is much better done in the native frameworks.
来源背景：Vincent Lussenburg，前 LinearB Senior Manager of Customer Solutions - 中性 开发者工具最终都在转向消费制定价
VL: Of what I'm able to disclose and what I'm able not to disclose. OK, so it is quite important and all the players are developing code review capabilities pretty quickly. So what I remember is that a lot of focus was put in... and this is public information... benchmarking, for example. So what LinearB focused on is creating a benchmark that compared all these different code review tools. And that was actually a decent benchmark; I kind of like that it's out there. I don't know if they're still developing it, but it would be able to show, OK, if I have a code base and this code base had certain problems in it, which of them, the AI tools, would actually find the problems in that. So comparing like a CodeRabbit and Copilot with LinearB's solution. Let's see. I think what I remember of the adoption is that some customers saw a little bit more of a positive signal in regards to Copilot, as in it was performing better than Copilot. In some situations, it was performing better than CodeRabbit, but not that much. But the way that I translate that is: Copilot, of course, the quality of the code review is partially driven by different models that you select because there, as a developer or as an organization, you can select the cheaper models or the more expensive models because GitHub ultimately started running a consumption-based enterprise. One of the reasons that LinearB was still able to get decent results from their code review was the reason that they used a relatively expensive model. So this is the typical situation what you see in the market a lot right now is that at first you're getting your tokens for free because it's a flat fee per contributor and of course everybody is switching to a consumption model eventually. So if there would be adoption, I guarantee you that ultimately those customers would end up paying for that. But that's one of the reasons that the code review would perform decently in comparison to competitors. And less so to competitors that really focused on code review as their primary only product. Like, for example, CodeRabbit was really focused on that as a niche. So LinearB has always been lagging behind what CodeRabbit did because that's sort of their main product. Yeah, that's what comes to mind, thinking for the rest what I remember about the AI code reviews.
来源背景：Vincent Lussenburg，前 LinearB Senior Manager of Customer Solutions - 反对 客户完全停用 Jira，改在 Notion 内做工单，成本减半
SG: The benefit, in terms of cost, I think for me, it's the main benefit because at first we had Notion that we have to pay plus the license on Jira, but we cut Jira totally. So we cut our cost by a factor of 2. Basically, in terms of costs, it's for me the main benefit. And also, the other one is that everything is embedded inside Notion. So you have your tickets, but you have also all your knowledge around the tickets.
来源背景：Sidney Golstein，Rakuten France Lead AI PM；Rakuten France 约 200 人 - 反对 客户认为用 Notion 可替代整个 Atlassian 套件，额外成本为零
SG: Okay, but basically we have a backlog for me, as far as I see it or I understand it, for me it's just some database. There is a database of tickets with different properties. For instance, we have also a database of feature teams. So every ticket will be related to 1 single feature team. So you have a database of tickets, a database of feature teams, you have a database of sprints, you have a database of release, you have a database of epics. And if you know how a backlog is working, you can connect all of these databases with exactly the same things and the same capabilities as in Jira. But the main advantage is that you have your backlog inside your whole documentation and your whole knowledge. So in terms of context, in terms of... For instance, we have also a database of PRD, product requirement documents inside Notion. So it's very easy to then create... what I have also built... an agent to create the tickets for the team. So the product manager, for instance, queries this agent. The agent has the whole knowledge of this feature team and the different projects they are working on. And the most important thing is the context and the access to this context. And if you have everything inside Notion, you can pretty much do... It's very powerful. You don't have to connect to Jira, you don't have to connect to Teams, you just have everything on Notion, and that's for me the... The way to create a very strong backlog and the main advantage is that it's free. The additional cost of creating a backlog is free, it's not more expensive, you can cut out your budget because you replace the whole Atlassian suite with Notion directly. So for me, it's very interesting.
来源背景：Sidney Golstein，Rakuten France Lead AI PM；Rakuten France 约 200 人 - 中性 Service Collection 主要从三线厂商（ManageEngine、Freshworks 等）抢份额，而不是 ServiceNow
I look at it as the baseline for success of any business to have a service model that works and Service Collection is among the leaders there for sure. Can they continue to grow 30%? They still have quite a bit of runway in their customer base. They weren't even targeting before. As long as they make investments in the enterprise sales force… They made a huge investment in their enterprise sales force around it last year. That's why you're seeing a payoff. I think you'll see that investment they made pay off even more this year as more and more of the sales folks are actually getting to the point where they're creating revenue. Service Management can grow quite a bit over the next couple of years, even without necessarily taking a ton of share from at least ServiceNow. I think they're taking share from a lot of, I'll call them, third-tier vendors, ManageEngine, the Avantis, even the Freshworks to a degree, I think that they're battling it out in the small markets, small business space quite a bit.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20 - 口径不齐 MCP 调用量季度增长 400%，而 Web/移动端的人工使用基本没变：agent 访问量增长远快于人工使用（这句原意更像在说服务成本，有歧义）
Yes. Look, Karl, I can add on from the gross margin in the COGS side. I would say it's a classic Atlassian story of being thoughtful and long-term thinking -- long-term orientation. We've talked for 3 years now around our AI gateway and how we think about AI. We've been blending multiple models for a multiyear period, and we're getting really, really good at it. It's a nontrivial challenge. I'll give a huge shout out to all of our R&D teams who have been working on that cost to serve question for a long time in terms of getting the best quality at the best price and the best speed. People want [ strike ] fast answers from their AI across Teamwork Graph, across our enterprise search architecture, which is now the best in the world in terms of giving you answers across your knowledge estate, whether that's to a person or human or whether that's to an agent or something clearing in that manner from an AI perspective. Our ability to blend models, our ability to manage traffic costs over time is truly phenomenal. We get better at it every quarter. And you can see that we had a huge continued growth in AI usage. Again, Rovo assisted actions are up 50% quarter-on-quarter. And we've continued to do that and focus on the durable profitable growth that we've talked about for a while now. So it's a huge credit to everyone in the Atlassian team to be able to do that. It's a lot of hard work. It's a lot of long-term thinking about the architecture that we built underneath our AI to be able to give those answers in a -- in an incredibly competitive value way to our customers as well. So not only do we give them the best answers, we give them the best answers at an incredible price, which is part of the reason you're seeing the growth in the service question, the growth in the Teamwork Collection, again, as James mentioned service question, again accelerating AI being a huge driver of that service space. Again, hearing record numbers for customer conversions from legacy service providers, et cetera. So -- and you see that with our MCP calls as well as an example, that's also nontrivial to serve cheaply to manage that over time. MCP calls were up 400% in the quarter with barely a blip in people actually using the applications through the web or through a mobile client. So really happy where we sit from a technological perspective of being able to deliver this and more improvements to come.
- 中性 MCP + CLI 用户季度突破 100 万 MAU，环比翻倍以上；agent 通过 MCP/CLI 访问 Teamwork Graph（没说是否按 seat 收费）
This is about being an interconnected part of a customer's technology estate. It's about the Atlassian platform being a strategic piece in the puzzle of how it is that they run their business. And I think that integration is a really important point. We're seeing that the reason we talk about the MCP server, MCP and CLI users in the quarter, passing 1 million MAU. We don't know where that puts us in a world scale, but I think it's 1 of the biggest MCP servers that exist, more than doubling in the quarter is because of people getting to the Teamwork Graph getting to that context from their agents wherever they're deployed. Again, we also signed and increased our partnerships with all of the Foundation Labs. During the quarter, we released Cursor and Claude and GitHub agents within Jira to again bring our interfaces, those agents into our interfaces, so you can use them as natively as possible so that we are part of the network of tools that help an integral part, the strategic part of the network of tools that helps those companies to do what they need to do.
- 中性 消费制：超额使用直接收费 + Flex（按金额承诺跨 collection 使用），都处在早期；TWC 和 Service Collection 是 AI 变现载体
Rob, look, as we said, we continue to see a great increase in Rovo assisted actions, which is probably a rough proxy for credit usage. It's not directly linear, but the 50% quarter-on-quarter growth gives you a sense of the customer adoption of Rovo across the areas. Secondly, as we've said that the Teamwork Collection with approximately 10x as many Rovo credits included is a great motivator inducement for upgrade in terms of hitting those limits and looking at upgrading the Teamwork Collection, both in terms of moving to the Teamwork Collection, but also expanding the number of seats, the reps of deployment. Again, as we talked about, as I see the Teamwork Graph working in most different types of teams, you see seat increases, right, which is, again, driving that ARPU number, if you're looking at it at a really macro level, that is happening there. We have a series of consumptive elements today. We continue to work on how we judiciously and prudently roll out consumption-based offerings. We talked about both the direct consumption-based offering for overages. And secondly, the Flex model, which we continue to work on to allow customers to opt in to the entire Atlassian platform with a sort of dollar-based commitment that they can use across various different collections and consumptive elements in a singular package, both of which are early in their cycle holds great promise for the future. But as always, we maintain our sort of long-term thinking position and working through there. Right now, the simplest thing I would say is that the Teamwork Collection and the Service Collection are the vehicles for AI monetization, right? Customers adopting those, driving the usage within those. That is as we've said, that's the vehicle to measure AI monetization and adoption and both are going very, very strongly.
- 反对 Rovo 超额未强制收费，消费计费 FY27 不贡献实质增长
Bears also note that direct AI monetization remains small. TEAM is prioritizing adoption, is not yet enforcing Rovo overages, and does not expect consumption billing to contribute meaningfully to FY27 growth. The higher ARR growth among Rovo users could partly reflect the fact that TEAM’s largest and most engaged customers were naturally the first adopters. Mgmt disputes that interpretation, but investors still lack a separately measurable AI revenue contribution. Additionally, TEAM itself expects enterprises to maintain several knowledge graphs, leaving room for Microsoft, ServiceNow, data platforms and AI-native tools to own substantial portions of enterprise context.
来源背景：TMTB 作者转述业绩会与 callback - 口径不齐 consumption/usage 定价会越来越重要；今年开始执行 Rovo credit 限额
I get 10x the amount of credits across my entire user base at a higher price point, so we realize higher ARPU as a result instead of worrying about Jason is going to chew through x amount of credits and I have to worry about this power user. It's a fungible pool of Rovo credits spread across my entire organization at 10x the amount. So it's much more predictable and customer-friendly, and we're just meeting customers where they're at today. I think over time, consumption-based pricing or usage-based pricing becomes a bigger part of the story. We'll actually begin to enforce Rovo credit limits this year. But again, I think the primary motion you're going to see is customers choosing to adopt Teamwork Collections. So I would continue to pay attention to that because that's probably the primary monetization motion today. Customers are also -- when they adopt Rovo, they, of course, realize significantly more value.
- 中性 专家对 Atlassian 进入 80% 财富 500 强的数据口径存疑
专家：这是一个很好的问题。我无法给出完整答案，但了解其中一部分。Atlassian 的数据表现非常好，Rovo 辅助操作量环比增长约50%，增幅非常显著。此外，Atlassian 已进入《财富》500强企业中的80%，不过这一数据可能通过统计口径进行修饰，使其看起来优于实际情况。
来源背景：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团；自称并非 Atlassian 专家 - 中性 agent 不按 seat 收费：简单 Teamwork Graph 查询免费，复杂/长周期任务消耗 credits；12 月开始执行
And we already have some of them in our CSM product. But I think the the I think advantage that Atlassian has as a horizontal platform is the ability to stand up different types of be able to tackle different types of workloads and work that happens across organizations. And right now we see that through to more collection in terms of the ability to monetize and folks you know choosing Atlassian to be able to deploy their AI the other way as I mentioned is through some of the CLP, Sealite and MCP use cases that we're seeing across the system as well. Now for simple queries into the teamwork graph, we don't charge for that. Right ? So if you're writing a ticket, if you're trying to resolve a quikticket, but if you're tapping into the teamwork graph with much more long horizon or complex use cases that draw on that teamwork graph, we do burndown those credits for customers. Again, it's building habits, like we said, across our customer base. first driving that adoption that that the vision that I'm being breaking out which as a reminder you remetering as of December. Yeah, that's right.
- 中性 分析师问 agent 数量将远超人类 seats，如何货币化；CFO 只回答 credits/MCP 数据，没谈 agent 是否算 seat
We talked a lot about seat count, but we actually talked about human seat count. So what I want to ask you is as more agenting systems take on the roles and personas of a human user to undertake tasks and touch systems, you know, they would necessarily need to knock on the door of Jira for tracking and workflow information and Confluence and Ben standpoint. So our it might be early days, but how are you thinking about agents touching your system and the teamwork graph and how are you monetizing that ? Because as far as the I can see, agents are going to be multiplicative if not orders of magnitude more in volume than human users and seats. So what's kind of the philosophy and thought process there and what are you seeing customers do there ? That's a great question. You know, when we think about Rovo and the TEAMWORK GRAPH and customers using agents and agent taking use cases across the Atlassian platform, we see a couple of things.
- 中性 约 10% Cloud 客户可能在 12/3 计费开始后面临用量收费，随 Rovo/agent 采用提升（agent 的变现走用量而非 seat）
Mizuho says TEAM’s latest Cloud increases imply ~7–10% list-price uplift, modestly above last year, while expanded usage-based pricing could become a more material FY27 contributor. Mizuho estimates ~10% of Cloud customers may face incremental usage charges when billing begins Dec. 3, with penetration rising as Rovo/agentic AI adoption scales and potentially lifting ARPU through higher-tier Teamwork Collection adoption. Mizuho reiterates Outperform and raises PT to $215 from $175, with TEAM remaining a top pick.
来源背景：Mizuho 卖方（TMTB 转述） - 口径不齐 Jira/Confluence 不应按开发者 seat 建模，正转向管理人类与 agent 的记录系统
从我个人了解到的情况来看，实际情况介于两者之间。AI 提效并没有导致研发人数（developer headcount）同比例缩减；更贴切的描述是工程团队人数保持相对平稳（relatively flat），处于低个位数增长（low single-digit growth），而产出增长得快得多。需要明确的是，生产力提升（productivity gain）并不会 1:1 转化为端到端交付提升（end-to-end delivery gain）。但对公司而言，AI 编程工具确实让单个开发者产出了更多代码、PR 和变更。很多代码现在都是 AI 写的，占比可达 60%、70%、80% 甚至 90%，但这反而极大地增加了下游工作量。公司现在面临更多的 code review、CI/CD 运行、安全扫描（security scanning）、测试、文档、工单（tickets）、部署以及可观测性需求。到了2026年，我认为大约 80% 到 90% 的代码都由 AI 生成，但同时也带来了许多工作流问题。 对于像 Atlassian 这样的厂商（Jira 和 Confluence），不能简单地将其收入按开发者席位（developer seat）来建模。实际上，Jira 和 Confluence 本身正在从单纯的人类开发者席位工具，转变为融合人类与 AI Agent 工作流的记录系统（system of record）。Jira 正在明显向管理人类与 AI Agent 的规划、治理和度量层（planning, governance, and measurement layer）推进。 而像 Ansible 这类工具则有所不同。AI 产生了更多软件工程基础设施变更后，公司反而需要更多的自动化（automation）、配置管理（configuration management）、策略执行（policy enforcement）和可重复部署（repeatable deployment）。
来源背景：Strategy Executive，美国头部电信运营商（前任职） - 中性 开发工具定价从人类 seat 转向 seat + agent + 用量
核心不在于有多少开发者，而在于有多少基础设施和工作流任务需要自动化。如果对公司做一个方向性的量化，我认为未来一到两年内，开发者员工人数（headcount）的年增长率可能只有0%到5%，但软件产出和变更量可能会增长15%到30%。因此，传统的纯按人头席位（per-seat）收费的开发者工具将面临压力。但那些能够将AI Agent、自动化、工作流规模、安全治理以及基于用量（consumption-based）进行商业化变现的平台，反而有机会推动单开发者支出（spend per developer）上升。因此，真正的变化并不是AI导致开发者变少、进而引发开发工具支出下降；而是AI带来单开发者产出提高和更多机器生成的工作，促使开发工具定价逐渐从单纯的人类席位转向席位加Agent及用量收费。对于Jira、Confluence、CI/CD自动化和DevSecOps来说，这一趋势至关重要。
来源背景：Strategy Executive，美国头部电信运营商（前任职） - 反对 企业将通过统一 agent 层精简独立 workflow、ticketing、knowledge search 工具，外围 SaaS 有 10-20% 精简空间（未点名 Atlassian）
我觉得未来确实会考虑，但公司的方向并不是自己把这种 system of record 重新做一遍，而是保留 SOR，对上层一些重复的 application 和 workflow spend 进行 consolidation。像 Salesforce、ServiceNow 已经掌握了 authoritative data、permission、workflow 和 API，公司完全可以在它们之上建立自己的 agent abstraction layer。用户未来不一定需要进入五六个不同的 SaaS UI，而是可以通过统一的 enterprise agent 发起请求，再由 agent 调用 Salesforce、ServiceNow 或其他 SOR 来完成 transaction。这样公司就可以逐步减少一些 standalone workflow、ticketing、knowledge search、basic automation 以及 thin application layer tools。 如果量化的话，我认为未来两到三年（maybe two to three years），这类外围 SaaS 即 point solution spend 大概有 10% 到 20% 的 rationalization opportunity，但并不一定会全部砍掉。省下来的资金也不一定会直接转化为 cost savings，其中一部分可能会重新投入到 SOR 的 AI module、agent platform、integration API、data layer 以及 security governance 中。因此最终可能会出现一个很有意思的现象：application 的 vendor count 下降，但 Salesforce、ServiceNow 等核心 SOR 的 wallet share 反而上升。对公司而言，目标并非单纯少花钱，而是从众多 human-facing applications 转向更少的 system of record 加上一个 common agent interaction layer。
来源背景：Strategy Executive，美国头部电信运营商（前任职） - 中性 TEAM 被市场归类为 seat 型软件，随空头回补上涨
+ **ACN +6%; CTSH +5.2%; IBM +2.8%** led the AI-loser squeeze, CTSH from -4% at the morning lows after Vance said the US is suspending MSFT and other firms from the PERM green-card program. Rotation took over from there, extending ACN’s run since its 10/1 print when it was up mid teens. + **HUBS +5.3%; MNDY +6.5%; FIG +4.7%; TEAM +4%; ADBE +3.6%**: seat-based software also squeezed higher helped by the semis drawdown and short-covering, HUBS even after JPM downgraded it this morning (AI monetization taking longer, a medium-term growth headwind rather than a temporary one). * **ORCL -5.6%** on OpenAI related weakness. The WSJ story last night had ORCL talking to Apollo and GS about a vehicle that would own chips and lease them back. That supports medium-term bear case that FY27 funding leans on others with correlated exposure to the biggest AI counterparties; bulls will say customer-owned hardware and supplier financing are exactly how capacity outgrows ORCL-funded capex. * **MSFT -1.3%** with the OpenAI read-through on top of Vance’s green-card headline. * **PLTR +2.4%** was up as much as 5% after GS upgraded to Buy, arguing sovereign AI, bespoke apps and verticalization drive another step-up in TAM and another estimate-revision cycle into 2027. Getting close to ATHs again. * **U +1%** faded for a second day from +4% on RJ’s upgrade; several sell-side notes out saying how GOOGL’s AI and distribution close the non-coder gap in the creator funnel and Jefferies’ buyer survey saying U the biggest expected ‘27 share gainer. * **APP -42bps**: Jefferies cut ‘27 EBITDA 7% as its survey has APP’s gaming share flattening at ~30% after years of gains, putting the burden on take rate. * Cyber weaker: **PANW -2%; CRWD -1%** although **OKTA +1%** continues to act well
来源背景：前 Atlassian Global Head of Enterprise Sales（In Practise 推文引述） - 中性 企业对 AI 支出越来越价格敏感（潜在压制 AI 提价）
* Cannon-Brookes says Jira and the Teamwork Collection just had their best year of growth, accelerating at massive scale in users. * Atlassian sells products, including via DX, showing customers token spend by model and feature and where to cut it; he says enterprises are increasingly price-sensitive as AI becomes a growing budget line. * In Amsterdam, Atlassian announced agent-to-agent integrations with Figma and Databricks, so Confluence users can query Databricks or Snowflake through the Teamwork Graph, making Atlassian a hub rather than a bypassed destination (our read). * Atlassian runs many fine-tuned open-weight models, including a lot of NVIDIA Nemotron, and his biggest change of mind this year is inference prices falling further than expected, which suggests margin upside on AI features (our read).
来源背景：Atlassian CEO Mike Cannon-Brookes（摘要转述）
Information Add-on
从原文数字再算一层。
| 项目 | 原文输入 | 算式 | 结果 | 对照 | 状态 |
|---|---|---|---|---|---|
| FY27 有机净新增 ARR | FY26 末 subscription ARR $6.6B (2026-08-06 L401)，增速 23%，其中约 1pt 来自 DX；FY27 指引 18% | FY25 末 = 6.6 ÷ 1.23 = $5.37B；FY26 净新增 = $1.23B，减 DX 约 $0.05B = $1.18B；FY27 = 6.6 × 18% = $1.19B | 有机净新增 ARR 持平 | E05/E06：指引里没有加速 | 换算所得 |
| TWC 占 Jira seat 多大 | Jira ARR $2.5B 以上；每 seat $7-8.48/月；TWC 100 万 seats (2026-02) | 2.5B ÷ (7×12) ≈ 2,980 万 seat 当量；按 $8.48 约 2,460 万 | TWC 约占 3-4% | E04：打包渗透很早期，空间大，但也解释不了过去一年的大部分加速 | 待验证推断 (ARR 含折扣和 DC) |
| TWC 按挂牌价的 ARR | 100 万 seats；TWC Standard $14.02/月 (官网) | 1M × 14.02 × 12 | 约 $1.7 亿，约占 $6.6B 的 2.5% | E04：其中真正的增量只是升级差价 | 待验证推断 (未计折扣和版本) |
| "快 5%"到底多少 | CEO："比其他组的扩张率高 5%" (2026-02-05 L441) | 若基准扩张率 10%：相对口径 → 10.5%；百分点口径 → 15% | 影响在 0.5-5 个点之间，取决于口径 | E03：只覆盖数千家软件团队客户 | 口径未明 |
| 非开发者占比漂移是否真实 | Jira 非技术用户 48% (2025-06-03 L160) → 65% (2026-05-27 L42) | 假设 Jira 用户整体 +10%：非开发者 (0.65/0.48)×1.1−1 = +49%；开发者 (0.35/0.52)×1.1−1 = −26% | 开发者一年少 26% 不合理 | E02：口径变了，不能当趋势用 | 换算所得 |
| 迁移红利的尾巴 | 迁移第一年约 1:1，第 2-3 年恢复全价 (2026-03-10 L204)；迁移后 3 年 ARR 约 1.75x (2026-05-06 L567)；DC 2029-03 停服 | 2029 年迁移的客户，提价和扩张持续到约 2032 年 | "一次性"其实是到 FY31-32 的长尾 | E05：迁移不是一两年就结束的一次性，但有终点 | 换算所得 |
| 股价修复幅度 | 周收盘 $314.28 (2025-02-03 周)、$57.15 (2026-04-06 周)、$214.75 (2026-10-19 周) | 1 − 57.15/314.28；214.75/57.15 | −82%；3.76x | E06：预期修复已经走完一大段 | 换算所得 |
| 两年一倍需要什么 | 现价 $215；primer bull $352、base $166 (2026-10-16 primer) | 215 × 2 = 430；430 ÷ 352 | 比 bull case 还高 22% | 用户标准：现价不满足 | 换算所得 |
完整数字核对
所有和 seats、用户、打包直接相关的公司披露数字。公司从未披露 seat 总数、seat 净增或 seat 增速。
| 日期 | 指标 | 数值 | 谁说的 / 出处 | 口径问题 |
|---|---|---|---|---|
| 2025-06-03 | Jira 用户结构 | 52% 技术 / 48% 非技术 | IR Lam，Baird 会议 L160 | "用户" |
| 2025-06-04 | seat 结构 | 50% 技术 / 50% 非技术 | IR Lam，William Blair 会议 L222 | "seats"，全产品 |
| 2025-08-07 | 核心产品用户结构 | 50% 业务用户 | CEO，FQ4 FY25 L91 | "核心产品用户" |
| 2025-08-07 | Cloud NRR | 120% | CEO，FQ4 FY25 L83 | 之后只说"120% 以上、在升" |
| 2025-10-30 | 用 AI 编码工具客户的 Jira paid seat 扩张 | 快 5%，项目多 20%+ | CEO，FQ1 FY26 L274 | 相对口径，样本量不披露 |
| 2025-10-30 | TWC 带来的用户增长 | 两位数百分比 | CEO，FQ1 FY26 L88 | 无基数 |
| 2026-02-05 | TWC | 1,000+ 客户，100 万+ seats | CEO，FQ2 FY26 L106 | 此后只重复"超过" |
| 2026-02-05 | seat 扩张速率 | 企业和 SMB 稳定 4-6 个季度 | CFO Binz，FQ2 FY26 L393 | 定性 |
| 2026-03-05 | TWC 客户扩张 | 20% | CRO Duffy，Morgan Stanley 会议 L279 | 和 IR 的"10 个百分点以上"说法不一 |
| 2026-03-10 | Atlassian 内部开发者效率 | 约 +25% | IR Lam，Cantor 会议 L318 | 同时放缓招聘 |
| 2026-04-30 | Service Collection 非 IT 客户 | 60% | CEO，FQ3 FY26 L287 | 2 月说"超过 2/3" |
| 2026-05-06 | 非开发者用户占比 | Confluence 7/10、JSM 超 3/4、Jira 约 2/3 | CEO，Investor Day L153 | 不是付费 seat 口径 |
| 2026-05-06 | 剔除迁移的云 seats | 只有图，无刻度 | CEO，Investor Day L133 | 无法核实 |
| 2026-06-10 | TWC 升级时的用户增加 | 约 +10% | IR Lam，Mizuho 会议 L131 | 含竞品工具整合 |
| 2026-06-10 / 09-09 | Jira 价格 | 约 $8 / 约 $7 每人每月 | IR L362 / CFO L170 | 两次说法不一 |
| 2026-08-06 | TWC 客户每实例 paid seats | 单品客户的 4-5 倍 | CFO Chuong，FQ4 FY26 L199 | 横截面，选择偏差 |
| 2026-08-06 | FY27 subscription ARR 指引 | 18% (FY26 23%，含约 1pt DX) | FQ4 FY26 L113 | 隐含有机净新增持平 |
| 2026-08-11 | MCP / CLI | 月活 100 万+；经 MCP 建的工作项 4 倍；MCP 客户 ARR 增速 2 倍 | IR Lam，KeyBanc 会议 L220 | 工作量，不是 seat |
Thesis 之外的重要信息
- $1M 以上客户超过 600 家，过去一年 +39%；$3M+ 客户 164 家 (+50%)，$5M+ 69 家 (+70%)。增长集中在大客户，但前 VP 说 $1M+ 客户翻倍主要是老客户合并分散的合同，不是新类型的客户
- Fortune 500 覆盖 85%，只占约 10% 的收入。企业端渗透是管理层讲得最多的长期空间，也是 seat 增长最可能的来源
- Rovo credits 用得越多，毛利率压力越大 (CFO，Citi L254)。用量变现起来之后，利润率不一定跟着涨
- 竞品：Asana、monday 前员工都说在开发团队里很难撬动 Jira；Notion 在知识库评估中排在 Confluence 前面，有一家 200 人的客户完全停用了 Jira
Well, we talk about having 350,000 paying customers. We know that that's the bottom of this kind of pyramid illustration. How about at the top end? So our $1 million customer base, which I believe we said a quarter and a bit ago, had passed 600 customers spending more than $1 million a year. That has grown sixfold in the last four years. In the last year alone, it grew at 39% year-on-year, that category. These customers have more than a 99% retention year- on-year, and they are almost all expanding. If we look at our $3 million cohort over the last four years, when we talk about the enterprise investments we made on the technology side, on Brian's side, that $3 million cohort has grown tenfold in the last four years, 54% year-on-year in the very last year alone. And we have quite a number now of those $10 million-plus ARR customers.
**BR** [00:43:29]: It's definitely customers that have been Atlassian customers for a while, and they're now expanding their usage of it and consolidating their usage of it. The way that Atlassian used to sell, again, was 10s and 20s and 30s here and there, the 300,000 customers aren't unique customers. There's 300,000, maybe there's and Team 1 through Team 200 at Nike that uses, I'll use Nike again, that uses Atlassian before, but they were all separate contracts. Now consolidated those contracts into one, consolidated into one administrative model allows you to get into the… Actually, you're on the docket for someone like a CIO to even recognise now. You're like, "Oh, it's not just several hundred teams spending a few thousand dollars a year, it's now one team spending USD 1m per year, and that's one of my top vendors I need to spend time with you, which gives us the ability to upsell and upgrade from where you were before." Yes, they're absolutely going up. It's existing customers, it's upselling those existing customers and consolidating their usage, all of those.
来源背景：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow。做市场出身，看不到 seat 数据；"seat 不降反升"的论据来自他在 ServiceNow 的经历；2025 下半年仍在 Atlassian，离职时间未披露；访谈日 2026-07-20
And just maybe to add to that, I do want to reiterate what Mike said that it's not an overnight thing. We've got sort of similar questions last quarter about Service Collection when we had announced the $1 billion ARR growing north of 30% year-on-year. Those are quarters and years of investment that we're making here. And if you take a look at the enterprise go-to-market motion that we're continuing to evolve here, we really are still early stages when you think about the fact that we're in 85% of the Fortune 500, but really that only represents roughly 10% or so of our total revenue. There's still a lot of headroom there. And sure some of these stats a little bit earlier, but our $3 million cohort, that's 164 customers growing 50% year-over-year. Our $5 million and greater cohort of customers, that's 69, and that's on 70% year- over-year. But again, on a base of $6.6 billion of ARR, it really just highlights how early we are here in terms of the opportunity.
SG: The benefit, in terms of cost, I think for me, it's the main benefit because at first we had Notion that we have to pay plus the license on Jira, but we cut Jira totally. So we cut our cost by a factor of 2. Basically, in terms of costs, it's for me the main benefit. And also, the other one is that everything is embedded inside Notion. So you have your tickets, but you have also all your knowledge around the tickets.
来源背景：Sidney Golstein，Rakuten France Lead AI PM；Rakuten France 约 200 人
置信度检查
| 来源 | 检查范围 | 身份与视角 | 影响判断的问题 | 采用边界 |
|---|---|---|---|---|
| 6 场业绩会 (S&P) | 全文 | CEO、CFO (Binz → Chuong)、IR | 只给定性，关键问题 (seat 是否放缓) 被绕开 | 可作主要证据 (只限"管理层怎么说") |
| 11 场卖方会议 + Investor Day (S&P) | 全文 | CEO、总裁 Anu、CRO Duffy、CFO、IR | 非开发者占比口径多次变化；CRO 和 IR 对同一指标说法不一 | 可作主要证据，数字需交叉核对 |
| Citi 会议 (FactSet 近实时转录) | 全文 | CFO | 转录质量差："See growth""suite growth"；$7 那句断句缺失 | 需交叉验证 |
| Third Bridge：Atlassian 前 VP | 全文 | 前 VP，近期在 ServiceNow | 核心论据来自 ServiceNow；数字自称举例；部分数字是访谈人转述公开材料 | 需交叉验证 |
| GuidePoint：金牌渠道商 | 全文 | 渠道商 CEO，主做国防 | 增长集中在单一国防客户；长期观点是推测 | 可作一手线索 (样本窄) |
| Funda、SmartBear、GitLab、Tegus | 全文或关键段落 | 客户高管、同类公司前员工 | 不是 Atlassian 数据，是同类模式的类比 | 仅作线索 |
| 微软专家电话 | 仅检索关键词段落 | 微软战略客户 CTO | Rovo 数字是转述公司口径 | 仅作线索 (未全文检查) |
| 竞品专家 (Asana、monday、Notion、Smartsheet 等) | 仅检索关键词段落 | 竞品前员工或客户 | 站在竞品立场 | 仅作线索 (未全文检查) |
| TMTB 卖方摘要 | 只看 TEAM 相关段落 | Substack 汇总卖方和管理层 | 二手转述；个别结论 (如"$49M 超预期全部来自 seat") 比管理层原话更强 | 需交叉验证 |
影响判断的问题
| 原话 | 问题类型 | 对结论的影响 |
|---|---|---|
| "expand their Jira seats 5% faster" (2026-02-05 L85) | 口径未明 + 选择偏差 | E03 唯一的数字证据打折 |
| "65% of Jira users are knowledge workers" vs "48% are nontechnical" | 口径变化 | E02 不能算趋势 |
| CRO "20% expansion" vs IR "by double or by more than 10 percentage points" | 篇间矛盾 | TWC 扩张幅度取低值 |
| "roughly $7 per……" vs "$8 per user per month" | 篇间矛盾 + 转录断句 | 价格按 $7-8 区间用 |
| "The entire ~$49M Cloud beat came from core Jira/Confluence seat expansion" (TMTB) | 推断过度 | 管理层原话是"两大驱动"，没说全部 |
| "MCP calls were up 400%... barely a blip in people actually using" (2026-08-06 L242) | 字面歧义 | 可能在说服务成本，冲击 E03 的力度要打折 |
| "half nonbusiness" (Investor Day L154) | 可能转录问题 | 应为 non-technology，不影响结论 |
附：来源与阅读记录
共 54 份材料：管理层 18 份 (17 份 S&P 纪要 + 1 份 FactSet Citi 转录)，其余为专家访谈、竞品访谈和卖方摘要。同一场 Investor Day 的 FactSet 版本只用于交叉核对，不重复计数。分级：A 核心 19 / B 实质 19 / C 顺带提及 11 / X 无关 5。证据 391 条，另有反方原话 71 条，全部经脚本逐字比对原文。
专家与外部来源背景
| 日期 | 来源 | 背景 | 分级 | 可信度 |
|---|---|---|---|---|
| 2026-10-12 | Third Bridge | 专家（MS） | X | 仅作局部线索 |
S&P 纪要来自你坚果云的 TEAM 文件夹；其余来自 Knowledge Base，路径相对于 Knowledge Base 根目录。
B 2025-05-01 · S&P 业绩会纪要 · FQ3 FY25 业绩会：paid seat expansion 好于预期；Teamwork Collection 刚发布，管理层称对 FY26 影响有限 12 条
文件：/tmp/claude-0/seats/mgmt/2025-05-01_Earnings_Call.txt
发言人：CEO Mike Cannon-Brookes、CFO Joe Binz（明示）及卖方分析师
阅读范围：全文 L1-L549 可信度：可作主要证据
CFO 称云收入超预期来自'好于预期的 paid seat expansion'和 DC 迁移，但未量化。BMO 追问 Teamwork Collection 打包折扣对'价格 vs 席位'的影响，管理层称 TWC 对 FY26 影响有限，长期增长逻辑是 Confluence/Loom attach、版本升级、wall-to-wall 带来的新增 seats 和 AI credits。DC 增长主要靠提价。
问题：'additions' 多处为 'editions' 的转录错误（L126、L405、L490）；paid seat expansion 只给了方向（好于预期），没有给速率或数字；TWC 刚发布，对 seats/ARPU 的说法均为预期
A 2025-05-29 · S&P 卖方会议纪要 · 总裁 Anu：AI 让更多非程序员参与创造软件，客户 license 仍在扩张，seat+消费计价是“价x量”双顺风 12 条
文件：/tmp/claude-0/seats/mgmt/2025-05-29_Company_Conference_Presentation.txt
发言人：Anutthara Bharadwaj，President（明示）；主持人为 Unknown Analyst（推测 Bernstein，依据 L286 'at AB, at AllianceBernstein'）
阅读范围：全文 L1-L576 可信度：需交叉验证
主持人直接把“copilot 让产品团队变少、seat-based 业务受损”抛给管理层。Anu 的回答全是定性说法：AI 让更多人参与创造、客户 license 继续扩张、高技术 SMB 和非技术 SMB 扩张率一致，并把消费计价定位成 seat 之外的额外价格杠杆。seat 增长没有任何量化数字。她也承认 JSM 的 AI 自动处理工单可以不用人工，这对服务台 agent seats 是潜在压缩。
问题：seat 扩张全部是定性说法，没有给数字或增速；“AI 用户 150 万”说的是 AI 功能用户，不是付费 seats；“超过一半的客户/使用量是非开发团队”在 L68 和 L78 口径不同（客户 vs 使用量）；主持人身份未披露，只能从 AB 经历推测是 Bernstein
A 2025-06-03 · S&P 卖方会议纪要 · IR Martin Lam：过去 4 个季度 seat 增长一直很稳，Jira 用户 48% 是非技术用户，科技创业公司 seat 扩张和其他行业一致 12 条
文件：/tmp/claude-0/seats/mgmt/2025-06-03_Company_Conference_Presentation.txt
发言人：Martin Lam，Head of IR（明示）；分析师 Rob Oliver，Baird
阅读范围：全文 L1-L381 可信度：需交叉验证
Lam 主动说过去 4 个季度 seat 增长“非常稳定、健康”，科技创业公司的 seat 扩张模式和其他行业 SMB 一致，但没有给增速。他给出 Jira 用户 52% 技术 / 48% 非技术的拆分，并把 seat 扩张列为多个增长杠杆中的一个（还有版本升级、交叉销售、定价）。消费计价被定位为在 per-seat 之外获取价值的方式。
问题：“seat 增长非常稳定”没有给增速或数字；52/48 是 Jira 用户口径，而且是此前披露的数据；L238 又说“50% of our users”，口径从 Jira 扩到了全部用户；FY27 收入 CAGR 目标被回避
A 2025-06-04 · S&P 卖方会议纪要 · CRO Duffy + IR Lam：AI 不会让 seats 下降，“继续看到绝对 seat 增长”，TWC 要求各产品 seat 数量一致 12 条
文件：/tmp/claude-0/seats/mgmt/2025-06-04_Company_Conference_Presentation.txt
发言人：Brian Duffy，CRO（明示；主持人在 L24 误称为 CFO）；Martin Lam，IR；分析师 Arjun Bhatia，William Blair
阅读范围：全文 L1-L319 可信度：需交叉验证
分析师直接问 AI 让开发效率提升 30% 对 Jira 的影响。Duffy 说 seat 不但不会降，反而因为项目变多、使用变多而增加（依据是自己团队的经验）。Lam 补充：50% 的 seats 是技术类，SMB 科技创业公司的 seat 扩张一年来都很稳定，“继续看到绝对 seat 增长”，喊得最响的 CEO 们所在公司 seat 也在扩张。都没有数字。TWC 是同一个 SKU、Jira 和 Confluence 同等数量授权，等于在机制上带来 Confluence 和 Loom 的 seat 补齐。
问题：主持人把 Duffy 称为 CFO（L24），他实际是 CRO；Lam 这里说“50% of our seats”，前一天（6/3）说的是“52% of Jira users”，口径不一致（seats vs 用户、全部 vs Jira）；Duffy“更多使用带来更多 seats”的依据是自己团队的经验，属于推断；分析师说“去年披露 1.5M MAU”（L258），时间说错了
A 2025-08-07 · S&P 业绩会纪要 · FQ4 FY25 业绩会：CEO 称没看到 code gen 对 seat 数据有任何影响；50% 核心产品用户为业务用户；paid seat expansion 速率环比稳定 19 条
文件：/tmp/claude-0/seats/mgmt/2025-08-07_Earnings_Call.txt
发言人：CEO Mike Cannon-Brookes、CFO Joe Binz（明示）及卖方分析师
阅读范围：全文 L1-L585 可信度：可作主要证据
Morgan Stanley 首问 code gen 对 seat expansion 的冲击，CEO 称没在任何数字里看到影响，接入 code gen 工具的客户业务用户增长更快（早期数据，未量化）。CFO 称 Q4 云超预期主要来自 paid seat expansion、cross-sell 和迁移，SMB 和企业的 paid seat expansion 速率环比稳定（是稳定，不是加速）。CEO 披露核心产品 50% 用户是业务用户；Mizuho 质疑企业里非技术员工渗透率低，CEO 只给了定性回答。
问题：'cogeneration'/'co-generating' 为 code generation 的转录错误；'geo instances' 应为 Jira instances；'seed expansion' 应为 seat expansion；'additions' 应为 editions；'业务用户增长更快'只是早期数据，没给数字或样本；50% 业务用户（core apps 口径）与后续季度'2/3 知识工作者'（Jira+Confluence 口径）口径不一致；迁移贡献 FY26 指引转录为 'mid- to single-digit range'，原意应为 mid single-digit
B 2025-09-10 · S&P 卖方会议纪要 · Anu + Duffy：非技术用户是增长最快的用户群，AI 正在压缩软件开发角色，变现靠上云、TWC 和版本升级 11 条
文件：/tmp/claude-0/seats/mgmt/2025-09-10_Company_Conference_Presentation.txt
发言人：Anutthara Bharadwaj，President（即将卸任，L57）；Brian Duffy，CRO；分析师 Kash Rangan，Goldman Sachs
阅读范围：全文 L1-L441 可信度：需交叉验证
Anu 承认市场在担心 seats，她的回应是 AI 让非技术人员也能创造软件。但她同时描述了开发流程被压缩：多个角色的工作集中到通才身上，协作单元从人变成 agent。Duffy 说 50% 用户在业务端，而且业务端是增长最快的用户群（没给数字），并把变现路径定为上云迁移、TWC 和版本升级。AI MAU 达到 230 万。
问题：“增长最快的用户群在业务端”没有量化；L115 Anu 说“well over half our business has been serving nontechnical users”（业务口径），和 Duffy 说的 50% users 口径不同；L220 '[ consumers ]' 和 L77 '[ mau ]' 有转录标记
A 2025-10-30 · S&P 业绩会纪要 · FQ1 FY26 业绩会：使用 code gen 工具的客户 Jira paid seats 扩张快 5%；TWC 带来两位数用户增长；云指引上调只反映迁移 13 条
文件：/tmp/claude-0/seats/mgmt/2025-10-30_Earnings_Call.txt
发言人：CEO Mike Cannon-Brookes、CFO Joe Binz（明示）及卖方分析师
阅读范围：全文 L1-L514 可信度：可作主要证据
首次给出 cohort 数据：使用 GitHub Copilot、Cursor 等 code gen 工具的客户 Jira paid seats 扩张率比未使用者快 5%，项目数多 20%+。Mizuho 追问 cohort 规模，CEO 没给。CEO 称 TWC 推动用户数两位数百分比增长。但 CFO 明确云指引上调 1.5pt 只反映迁移，其他有机驱动没上调，剔除 DC EOL 后的有机收入指引反而下调 0.5pt。
问题：'5% faster' 是相对差还是百分点差，本篇没说清（FQ2 FY26 业绩会才澄清为相对差）；cohort 存在选择偏差（使用 code gen 的多为前沿、增长型公司），管理层拒绝披露样本规模；'migrations -- seats that were upgraded' 是迁移 seat，不是净新增 seat；'additions' 应为 editions；'cogeneration' 应为 code generation
X 2026-01-16 · MeritForum · NOW 4Q 调研：仅提到 Now Assist 推广需要与 JSM 集成 0 条
文件：Knowledge Base/MeritForum/2026-01/2026-01-16-now-4q26ye-ji-diao-yan-ying-wen.md
发言人：ServiceNow 渠道/专家（推测）
阅读范围：全文检索 Jira/Atlassian/seat，仅 L262 命中 可信度：仅作局部线索
只提到 POC 转为付费推广需要与 Jira Service Management 和 CSM 集成，与 TEAM seats 无关。
C 2026-01-29 · VIC · VIC ServiceNow 多头：反驳 'seat 收入会下降' 论，作者此前写过 Atlassian 多头 3 条
文件：Knowledge Base/VIC/2026-01/2026-01-29-servicenow-inc.md
发言人：VIC 作者 jgalt（买方）及其引用的 X 帖
阅读范围：全文检索 Atlassian/Jira/seat 段落 L40-L82 可信度：仅作局部线索
作者用自己的 Atlassian 多头文章反驳 AI 导致 seat 压缩的空头论点；引用的反驳称 AI 提效使企业招更多人（更多 seat），并转向消费/结果定价；还称 Atlassian 在 Jira 里加 AI coding agent 比创业公司容易。都是一般性论证，无数据。
问题：一般性论证，无 TEAM 数据
A 2026-02-05 · S&P 业绩会纪要 · FQ2 FY26 业绩会：TWC 售出 100 万+ seats；seat 扩张跨技术/非技术用户、速率稳定 4-6 个季度；JSM 承认部分客户出现效率化 20 条
文件：/tmp/claude-0/seats/mgmt/2026-02-05_Earnings_Call.txt
发言人：CEO Mike Cannon-Brookes、CFO Joe Binz（明示）及卖方分析师
阅读范围：全文 L1-L557 可信度：可作主要证据
软件股暴跌背景下（分析师称股价 5 周跌约 40%），管理层重申 seat 模型：使用 code gen 工具的客户 Jira seats 扩张快 5%（CEO 澄清是相对快 5%，不是 5% 的扩张率，并承认这些多是前沿增长型公司）；TWC 不到 9 个月售出 100 万+ seats，相比单品有两位数 seat 扩张，但部分来自打包时各产品席位数对齐。CFO 称 seat 扩张跨技术和非技术用户，速率在企业和 SMB 稳定了 4-6 个季度，但没给拆分。BMO 问 JSM 同客户可比口径下 seat 是否萎缩，CEO 承认'部分客户出现效率化'。
问题：'5 million Rovo seats'（L192）实为 Rovo MAU，CEO 把 AI 使用者叫作 seats，口径混用；Service Collection 覆盖 50% Fortune 500（本篇 L487）vs 下季度 75%（FQ3 FY26 L287），需交叉核对；seat 扩张没有拆分为新客户/存量/技术/非技术，只有定性；TWC 两位数 seat 扩张包含打包时席位对齐的机械效应
A 2026-03-05 · S&P 卖方会议纪要 · CRO Duffy：续约时没看到 seat 减少，用 coding 工具的客户 seat 多 5%，TWC 客户扩张 20%，agent 怎么变现还在讨论 14 条
文件：/tmp/claude-0/seats/mgmt/2026-03-05_Company_Conference_Presentation.txt
发言人：Brian Duffy，CRO（明示）；分析师 Keith Weiss，Morgan Stanley
阅读范围：全文 L1-L429 可信度：可作主要证据
在软件股大跌的背景下，Weiss 直接问 seat 模型和 Jevons 悖论。Duffy 说客户对话里和续约时都没看到 seat 减少，用 coding 工具的客户 seat 多 5%，TWC 客户扩张 20%（举例：一家半导体公司 4 万 seats 上线 10 天后再加 8,000）。业务端是增长最快的部分。agent 怎么变现还没定，“我们和其他公司都在讨论”。
问题：“5% increase in seats”没说基准（相对不用工具的客户？哪个时间段？）；L171 'a 5% increase in fees' 疑为转录错误（应为 seats）；“over 1 million seats in the cloud”放在 TWC 语境里，按 3/10 Lam 的说法应是 TWC 售出 100 万 seats，不是云端总 seats；TWC 扩张“20%”和 3/10 Lam 说的“多 10 个百分点以上”口径不同；销售人员：2025-06 说“目前约 400 人”，这里说“今年底约 400 人，较去年翻倍”，前后矛盾；“业务端增长最快”仍然没有量化
A 2026-03-10 · S&P 卖方会议纪要 · IR Lam：用 Cursor/Claude Code 的客户 seat 扩张快 5%，AI 50 中 60% 是客户，TWC 客户 seat 多扩张 10pp+，DC 迁移提价 0-2 倍 16 条
文件：/tmp/claude-0/seats/mgmt/2026-03-10_Company_Conference_Presentation.txt
发言人：Martin Lam，Head of IR（明示）；分析师 Thomas Blakey，Cantor Fitzgerald；另有现场 Unknown Analyst
阅读范围：全文 L1-L398 可信度：可作主要证据
这是 seat 量化最多的一篇。通过 MCP 用 Cursor/Claude Code 的数千家客户：Jira 任务多 5%、MAU 多 5%、seat 扩张快 5%。Forbes AI 50 中 60% 是客户，seat 增速远高于公司整体。企业客户在扩大 seat 数、签更长的合约；TWC 有 1,000 家客户、100 万 seats，seat 扩张比单买 Jira 多 10pp 以上。DC 迁移提价从持平/5% 到 2 倍，第一年约 1:1，第 2-3 年逐步回到全价。同时披露公司内部开发者效率提升约 25%，招聘在放缓。
问题：L188-190 的提问是 L37-39 的重复，疑似转录错误，对应回答实际是 DC 提价；“expand their seats 5% faster”（相对增速）和 3/5 Duffy 说的“5% increase in seats”口径不一致；TWC“by double or by more than 10 percentage points”表述含糊，和 Duffy 的“20% expansion”口径不同；用 coding 工具的客户可能本身就是增长更快的科技公司，有选择偏差，管理层没有控制这一点；AI 50 的“much faster”没有给数字；L326 'Does the seat growth even still that nobody wants to talk about' 转录有缺损
A 2026-04-17 · GuidePoint · Atlassian 金牌渠道商：PubSec seat 暴增、非 PubSec 符合预期；但长期看 AI 让多 seat 无必要，per-seat 将转向用量 14 条
文件：Knowledge Base/GuidePoint/2026-04/2026-04-17-atlassian-github-and-gitlab-april-2026-gold-partner-channel-check.md
发言人：Shawn Doyle，ReleaseTEAM CEO（自 1999 年），Atlassian Gold Partner，客户多为 Fortune 1000，重点 PubSec/国防（L40, L160-163 明示）
阅读范围：全文 L1-L488 可信度：可作主要证据
渠道一手数据显示 2026 Q1 seat 增长强劲，但主要来自 PubSec/国防（单客户 300→5,000 seat 申请，专家自己怀疑能否配齐人），非 PubSec 季度‘平淡、符合预期’；当下 AI 让客户增加资深开发者以审核 AI 代码，故‘会卖更多 seat’。但同一专家明确认为 AI 编码工具通过 MCP/服务账号接入 Jira 后‘不需要多 seat’，2-3 年内 Atlassian 必须转向用量计费，且到 AI 采用‘钟形曲线’顶点后会出现开发者大裁员；Service Collection 增长主要因 JSM 被强制打包续约。
问题：seat 增长主体是 PubSec/国防，单一事件驱动，不代表 Atlassian 整体；‘30-35% 增长’是该渠道商自身 Atlassian 业务，不是 Atlassian 公司增速；专家对当下（卖更多 seat）与 2-3 年后（不需要多 seat、裁员）的判断方向相反，需按时间维度区分；渠道商利益：用量计费会削弱渠道利润，其关注点带有立场
B 2026-04-26 · Tegus · Flock Safety IT 工程师：用 AI 原生 ITSM Console 叠在 JSM 之上，少招 4 名支持人员，计划用 Console 替代部分团队的 JSM 6 条
文件：Knowledge Base/Tegus/2026-04/2026-04-26-console-senior-it-systems-engineer-at-flock-safety.md
发言人：Senior IT Systems Engineer at Flock Safety，Console 客户（L91 明示）
阅读范围：全文 L1-L347（网页导出，含 Tegus 界面噪声），精读 L100-L300 可信度：仅作局部线索
终端客户实例：原本只用 JSM 做 IT 服务台，引入 AI 原生 Console（按全员 $10-15/人/月计费）后 30-40% 工单被拦截，原计划招 4-6 名支持人员只招了 2 名，即 JSM agent seat 增长被 AI 压缩；目前 Jira 仍是核心工单系统，但计划把 Console 作为一站式入口并扩展到财务、HR、采购等团队，专家认为 Console 能为部分团队完全替代 JSM。对 JSM seat 和‘非技术部门扩张’是反向证据（扩张给了竞品）。
问题：单一公司样本（快速成长的中型公司），代表性有限；JSM 按 agent 收费，‘少招 4 人’对应 seat 影响为推断，专家未提 JSM seat/合同变化；专家对 Console 偏正面（NPS 7.5-8）
C 2026-04-29 · GuidePoint · Canadian Tire PM：选择 Jira Align（企业级），定价按企业协议打包但仍间接随 seat 数扩展 2 条
文件：Knowledge Base/GuidePoint/2026-04/2026-04-29-project-portfolio-management-productboard-customer-highlights-efficiency-gains-in-product-discovery-but-limits-in-enterp.md
发言人：Mahak Varshney，Canadian Tire Product Manager（明示）
阅读范围：全文检索 Jira/Atlassian/seat/pricing 段落 L250-L340 可信度：仅作局部线索
客户在 Planview/Jira Align/Smartsheet 中选择 Jira Align 企业版，原因是已嵌入 Jira 生态降低采用摩擦；Jira Align 定价为企业订阅、与 Atlassian 协议打包，但仍间接随 seat 数扩展。
A 2026-04-30 · S&P 业绩会纪要 · FQ3 FY26 业绩会：云加速到 29%，CFO 称超预期来自 cross-sell 和 seat expansion；CEO 称没看到任何 seat 压缩信号；DC 迁移中客户的 seat 扩张放缓 17 条
文件：/tmp/claude-0/seats/mgmt/2026-04-30_Earnings_Call.txt
发言人：CEO Mike Cannon-Brookes、新任 CFO James Chuong（明示）及卖方分析师
阅读范围：全文 L1-L485 可信度：可作主要证据
云收入加速到 29%，CFO 称迁移符合预期、不是超预期来源，真正驱动是 cross-sell（TWC/Service Collection）和 seat expansion（包括 Jira 单品的持续 seat 扩张），均未量化。Mizuho 问 seat 增长驱动和持续性，CEO 称'没看到任何 seat 压缩信号，反而相反'，并称在以越来越快的速度扩展到非技术岗位。不利信息：正在计划迁移的 DC 客户 seat 扩张放缓，DC 收入超预期约 5000 万美元来自提价催化的提前采购。
问题：seat expansion 仍未量化，只说是超预期驱动之一；Rovo 客户 ARR 增速 2 倍可能有选择偏差（大客户、活跃客户先用）；Q3 总收入 32% 含 DC 提前确认的约 5000 万美元，不可外推；Service Collection 非 IT 占比：本季'60%'vs 上季'2/3 以上'；F500 覆盖：本季 75% vs 上季 50%，口径不一致；'Claude Pro' 可能为转录，原意或为 Claude；'service suction/question' 为 Service Collection 的转录错误
A 2026-05-04 · TMTB (Substack) · TEAM F3Q26 业绩点评：+20%，管理层称看到与 seat compression 相反的情况 7 条
文件：Knowledge Base/研究邮件/2026-05-04/TMTB Morning Wrap.md
发言人：TMTB 作者转述管理层与 callback；推测基于业绩会+卖方回访
阅读范围：仅检索 TEAM 段落 L203-L227（其余为其他公司） 可信度：需交叉验证
F3Q26（2026-04 季）TEAM Cloud +28.6% 大幅超预期，作者称管理层把超预期归因于 seat expansion + cross-sell 而非一次性迁移，并称看到 seat compression 的'相反'情况，NRR>120%。但同时指出 DC 客户在迁移前放缓 seat expansion、DC 约 $50M 提前确认、cRPO 减速；空头认为 AI agent 长期压缩 seat 的争论未结束。
问题：seat 增长未量化：没有付费 seat 数或增速，仅'seat expansion'定性；汇编作者转述，非业绩会原文
A 2026-05-06 · S&P Investor Day 纪要 · Atlassian 2026 Investor Forum：管理层集中反驳'seats 会消失'，展示剔除迁移的 cloud seats 曲线与非开发者用户占比 37 条
文件：/tmp/claude-0/seats/mgmt/2026-05-06_Analyst_Investor_Day.txt
发言人：Michael Cannon-Brookes（CEO）、Brian Duffy（CRO）、James Chuong（CFO）、Martin Lam（IR）；客户嘉宾 Cisco（Jason Andrews）与 Canal Plus（Stephane Baumier）；分析师来自 Mizuho/Bernstein/Baird/Canaccord/Raymond James/BofA/BMO
阅读范围：全文 L1-L1503 可信度：可作主要证据（管理层一手表述；但 seats 相关部分均为定性，数字需与 10-Q/股东信交叉验证）
CEO 开场即点名'seat 增长会死'的市场担忧，用剔除迁移的 cloud seats 图称每季度每年增长且未放缓，并披露非开发者用户占比（Confluence 7/10、JSM >3/4、Jira ~2/3），称增长早已不是开发者驱动；但 seat 增长全程未量化（无数量、无增速）。TWC 被定为主要 AI 货币化载体（>1,000 客户、100 万 seats、主要为存量升级、seat 数不变但 credits 10x），说明其贡献在 ARPU 而非 seat 数；同时管理层预期 FY27 总收入增速见底、DC 负增长，放弃 20% CAGR 目标，改用 ARR 加速作为增长指标。Q&A 中没有分析师直接追问 seat 数或 agent 是否算 seat。
问题：seat 增长仅以无刻度图表呈现（'get out your rulers'），未披露 seat 数、净增或增速——'seat expansion' 未被量化；L154-155 'half technology, half nonbusiness' 疑为口误/转录问题（应为 half non-technology），FactSet 版同样写作 'half non-business'；TWC 的 '1 million seats' 是升级口径（同等 seat 数换包），不是新增 seats，不应计入 seat 增长；Rovo/MCP 用户 ARR 增速 2x 为相关性数据，存在自选择偏差（CEO 对 credits 消耗 2x 也自承 'they've got more credits'）；agent 是否算 seat/是否收费未被直接说明；管理层只说多数 app 按 seat 定价，并在探索 flex 承诺制（可在 seats 与 apps 间挪用）；无分析师就 seats 直接提问；FY27 DC 下滑幅度 CFO 回避量化，延至 8 月指引；非开发者占比是用户口径（users），未说明是付费 seats 还是 MAU，也未给出该占比随时间变化的趋势
C 2026-05-07 · TMTB (Substack) · William Blair（Team '26 后）：迁移推动高价 SKU 和多年 Cloud 扩张 1 条
文件：Knowledge Base/Substack/2026-05/2026-05-07-tmtb-morning-wrap.md
发言人：William Blair（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L441-L448 可信度：仅作局部线索
William Blair 认为迁移、定价和基础设施优化提高利润率；DC 收入到 2027 年下降约 30%，迁移带动 premium/enterprise SKU 采用；无 seat 说法。
问题：说明 ARPU 部分来自迁移升档，偏一次性（与 E05 相关）
B 2026-05-08 · Third Bridge · SmartBear 前 CMO：seat 定价是‘dying thing’；客户裁员时 license 下降拖累 GRR；工作量回升可降低 seat 敏感度 7 条
文件：Knowledge Base/ThirdBridge/2026-05/2026-05-08-smartbear-from-tool-portfolio-to-integrated-platform-the-haloai-era.md
发言人：Cynthia Gumbert，SmartBear 前 CMO（2019-2025 初，L16-17, L23 明示）
阅读范围：全文 L1-L215（L113 以后为同一访谈的重复导出，引用取首次出现行号） 可信度：仅作局部线索
SmartBear（Zephyr 等 Jira 生态测试管理工具）前 CMO 证实：大客户裁员时 seat license 下降，对 GRR 有下行压力，下调主要表现为 seat 数减少而非降档；内部已认为 seat 定价‘正在消亡’，需要转向规模/用量/混合定价。QA 团队在缩小但工作量扩大，seat 持有人从 QA 转向开发者和产品负责人。对 Atlassian 是同类开发工具的旁证，偏负面；数据截至 2025 年初。
问题：对象是 SmartBear 而非 Atlassian，仅为同类旁证；专家 2025 年初离职，seat 观察可能早于 AI coding 大规模普及；裁员导致 seat 下降不一定由 AI 引起（含宏观裁员）
B 2026-05-12 · GuidePoint · 前 Asana APAC 销售负责人：Jira 在工程端根深蒂固且便宜；seat 模式在 AI 下有被压缩风险 4 条
文件：Knowledge Base/GuidePoint/2026-05/2026-05-12-asana-former-head-of-apac-corporate-sales-on-enterprise-adoption-ai-workflows-and-its-competitive-positioning.md
发言人：Robbie Goldberg，前 Asana APAC Corporate Sales 负责人（明示）
阅读范围：全文检索 Jira/Atlassian/seat 段落 L75-L82、L294-L326、L430-L450 可信度：仅作局部线索
专家称 Asana 难以打入工程团队，因为 Jira 在工程社区根深蒂固，$7/用户/月即可覆盖 80% 功能（vs Asana $33）；但对整个 seat 型协作软件，他指出若 AI 让 2 人完成 5 人工作，seat 型厂商会丢掉 3 个 seat，且用户可能直接用 Claude 而不进入工具。
问题：seat 风险论述针对 Asana，对 TEAM 是类比推断；L445 提到 'Atlassian, Microsoft, Asana, Wrike' 与 NRR 90s 的关系语义不清，可能转录问题，未采用
B 2026-05-19 · TMTB (Substack) · UBS：SaaS 是 2026 年 IT 预算输家，Atlassian 属最常被降优先级的类别之一，seat 优化成为新逆风 1 条
文件：Knowledge Base/Substack/2026-05/2026-05-19-tmtb-morning-wrap.md
发言人：UBS 卖方（TMTB 转述）
阅读范围：仅检索 Atlassian 段落 L180-L188 及开头 L16 可信度：需交叉验证
UBS 企业软件调研称客户把预算优先给 AI/云/安全/数据，seat 型应用软件被降优先级，Atlassian 被点名；合同期缩短、价格抵抗、seat optimization 成为 cRPO/bookings 新逆风。与 thesis 相反的需求侧调研证据（时间在 F4Q 前）。
问题：行业调研，未给 Atlassian 具体数据
B 2026-05-19 · GuidePoint · 前 monday.com 区域 PMO 总监：扩张从'买更多 seat'转为'seat+用量'；纯 per-seat 厂商要么转用量要么消亡；Monday Dev 打不过 Jira 4 条
文件：Knowledge Base/GuidePoint/2026-05/2026-05-19-monday-com-former-regional-pmo-director-on-ai-work-platform-evolution-and-enterprise-workflow-expansion.md
发言人：Jessica Saranich，前 monday.com Regional PMO Director & Chief of Staff（明示）
阅读范围：全文检索 Jira/Atlassian/seat/usage 段落 L238-L255、L300-L315、L335-L370、L392-L416 可信度：仅作局部线索
专家认为过去 10 年 SaaS 扩张靠买更多 seat，现在变成 seat+用量，扩张可在已有部门内通过用量加深；若 agent 能做 20 人的工作，per-seat 厂商必须转向用量否则会消亡。对 Jira：Monday Dev 很弱，纯开发团队里 Jira 会赢。
问题：专家观点针对行业，对 Atlassian 是推断
C 2026-05-20 · GuidePoint · 现任 Atlassian 解决方案工程师（前 Smartsheet）：Smartsheet 知识图谱落后 Atlassian Teamwork Graph 两年以上 1 条
文件：Knowledge Base/GuidePoint/2026-05/2026-05-20-smartsheet-former-enterprise-solutions-engineer-on-its-product-portfolio-and-ai-led-platform-evolution.md
发言人：Nordin Abajtit，现任 Atlassian solutions engineer、前 Smartsheet（明示，可能偏向 Atlassian）
阅读范围：全文检索 Atlassian/agent 段落 L142-L150、L244-L262 可信度：仅作局部线索
专家认为 Smartsheet 的 Smart Hub/知识图谱是追赶，Atlassian Rovo Studio 和 Teamwork Graph（2023 年 4 月发布）领先两年以上；未谈 seats。
问题：专家现受雇于 Atlassian，存在利益相关
C 2026-05-21 · GuidePoint · 前 monday.com Academy 经理：Monday Dev 难以从 Jira 转化大型开发团队；看好 seat+消费混合定价 2 条
文件：Knowledge Base/GuidePoint/2026-05/2026-05-21-monday-com-a-former-customer-academy-manager-on-its-product-portfolio-and-ai-work-platform-strategy.md
发言人：Julia Fagelman，前 monday.com Customer Academy Manager（明示）
阅读范围：全文检索 Jira/seat/consumption 段落 L228-L240、L285-L306、L418-L437 可信度：仅作局部线索
Monday Dev 适合小团队，难以转化用 Jira 的大型复杂开发团队；认为 seat+consumption 混合定价可成为收入驱动。对 Jira seat 挤占无直接证据（反而显示大企业 Jira 防守稳）。
B 2026-05-27 · S&P 卖方会议纪要 · 新任 CFO James Chuong 在 Jefferies 会议：非开发者占比、TWC 100 万 seats、FY27 收入低谷 9 条
文件：/tmp/claude-0/seats/mgmt/2026-05-27_Company_Conference_Presentation.txt
发言人：James Chuong，CFO（明示）；主持 Brent Thill (Jefferies)
阅读范围：全文 L1-L331 可信度：可作主要证据
CFO 给出 Jira/Confluence/JSM 用户中非开发者占 65%/70%/75%，TWC 超 1,000 家客户、超 100 万 seats，并把 AI 降低编码门槛→更多软件→更多规划/追踪需求作为 seats 逻辑；但没有量化整体 seat 增速。同时承认正准备迁移上云的客户扩张'muted'，FY27 会出现机械性收入低谷，并提到 AI 货币化未来可能更多转向 consumption '与 seats 并行'。
问题：未量化整体 paid seat 净增/增速，只有 TWC seats 存量；'non-edge, non-dev' 可能是 non-eng 的转录；$3M+ 客户 54% 增速与 8 月/9 月口径（50%）略不同
B 2026-06-01 · Third Bridge · GitLab 前东区销售总监：seat 缩减已普遍发生，一年后 seat 定价将失去意义；Atlassian 不构成竞争 9 条
文件：Knowledge Base/ThirdBridge/2026-06/2026-06-01-gitlab-usd-1bn-arr-inflection-point-sales-strategy-ai-monetisation-competitive-resilience.md
发言人：Steve Xu，GitLab 前 Eastern US 销售总监（2019-2025），现就职 dbt（L17-18, L26 明示）
阅读范围：全文 L1-L264（L196 以后为重复导出，引用取首次出现行号） 可信度：需交叉验证
作为同类 per-seat 开发工具对照：前 GitLab 销售总监称在 GitLab 和现职公司都看到 seat 缩减，客户把预算转向 Anthropic/OpenAI，开发者变少导致 seat 预算减少，必须转向用量；NRR 下滑主要来自中小客户，续约需更多折扣，AI 让迁移更容易、黏性下降。对 Atlassian thesis 是强反面对照，但专家对 Atlassian 本身只说其仅做项目管理。
问题：对象为 GitLab，仅作同类对照；GitLab 客户集中于开发者，Atlassian 有更多非开发者 seat；专家 2025 年离职，现职 dbt，部分观察来自非 GitLab 场景；专家持有 GitLab 股票（L30 自述），立场偏负面但并非看空动机；同一渠道商（GuidePoint 文）称 GitLab 非 PubSec 未见 seat 减少，与本文冲突
C 2026-06-01 · Third Bridge · 前 Symphony Talent 产品负责人：知识库评估中 Notion 胜过已有的 Confluence 1 条
文件：Knowledge Base/ThirdBridge/2026-06/2026-06-01-ai-workflow-automation-market-ema-unlimited-former-senior-product-role-at-symphony-talent-llc.md
发言人：前 Symphony Talent 高级产品岗位（明示）
阅读范围：全文检索 Confluence/Jira 段落 L20-L43 可信度：仅作局部线索
公司已有 Confluence，评估 Notion/Guru/Confluence 作为知识库，专家称 Confluence 老旧、数据糟糕、agent 推进慢，Notion 排第一。竞品挤占 Confluence 的单点证据。
问题：未明确是否已经削减 Confluence 席位
A 2026-06-02 · S&P 卖方会议纪要 · CFO 在 BofA 会议：Q3 Jira 独立 seats 超预期（开发者+业务 seats 均有），正面回应 seats bear case 8 条
文件：/tmp/claude-0/seats/mgmt/2026-06-02_Company_Conference_Presentation.txt
发言人：James Chuong，CFO（明示）；主持 Koji Ikeda (BofA)
阅读范围：全文 L1-L325 可信度：可作主要证据
分析师直接以 bear case 问 AI 对 seats 的长期影响，CFO 回答：Q3 TWC 与 Jira 独立 seats 均超预期，且开发者/工程 seats 与业务 seats 都在增长；AI 降低编码门槛→更多软件→更多协作需求。仍是定性说法，没有给出 seat 净增数字。另给 cloud NRR >120%，并称商业模式会在 seat 与 consumption 之间演进。
问题：seat 超预期没有量化（无 seat 数、增速）；'Jira stand-alone seats outperformed' 是相对内部预期的超出，不等于绝对增速很高
B 2026-06-09 · TMTB (Substack) · Wells Fargo：seat 增长改善 + TWC 升级 ARPU 超预期，PT $140 3 条
文件：Knowledge Base/Substack/2026-06/2026-06-09-tmtb-morning-wrap.md
发言人：Wells Fargo 卖方（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L295-L301 可信度：需交叉验证
Wells 认为 3Q 强劲由 Jira seat 增长和 TWC 升级驱动，TWC 的 ARPU 提升大于最初预期；投资者对 AI 的担忧从逆风转为潜在增长驱动。
问题：无数字
A 2026-06-10 · S&P 卖方会议纪要 · IR Martin Lam 在 Mizuho 会议：剔除迁移后 seat 增长持续复利；TWC 客户用户数多 10%；Jira 约 $8/用户/月 11 条
文件：/tmp/claude-0/seats/mgmt/2026-06-10_Company_Conference_Presentation.txt
发言人：Martin Lam，Head of IR（明示）；主持 Gregg Moskowitz (Mizuho)
阅读范围：全文 L1-L402 可信度：可作主要证据
IR 称 Q3 cloud 超预期两大驱动是 TWC 交叉销售和核心 Jira seat 扩张，并提到投资者论坛展示的'剔除迁移后 seat 增长持续复利'图表（本稿无数字）；客户升级 TWC 时用户数约多 10%（整合其他点方案）；AI 时代人与 agent 的协调需要更多人加入平台。另披露 93% 的 DC 迁移客户进入高版本，cloud 中加人更容易，以及 Jira 约 $8/用户/月的低价定位。
问题：'seat growth ex migrations' 图表数字不在本稿中，需找投资者论坛材料；L76 把 Teamwork Graph 的数字说反了（48% 更好/44% 更少 token），其他场合是 44%/48%，口误或转录问题；Jira $8/用户/月 与 Citi 会议 CFO 说的约 $7 不一致（可能口径不同：list vs 实收、或含/不含提价）
B 2026-07-01 · TMTB (Substack) · Citizens 渠道调研：18 个数据点 11 个负面，Jira 向 Linear 丢份额 2 条
文件：Knowledge Base/Substack/2026-07/2026-07-01-tmtb-morning-wrap.md
发言人：Citizens 卖方（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L232-L239 可信度：需交叉验证
Citizens 维持 Market Perform，渠道调研 11/18 负面，主要是 Jira 向 Linear 丢份额，DC 收缩为最大结构性逆风；承认 Rovo 和 Collections 有进展。是少数直接与 seat 增长 thesis 相悖的卖方渠道证据。
问题：未说明份额流失的规模、客户类型（可能偏中小/初创）
B 2026-07-08 · TMTB (Substack) · KeyBanc 下调 PT 至 $115：下调 FY27 Cloud 的迁移和有机增长假设；M-sci 称新增客户 logo 环比小幅改善 2 条
文件：Knowledge Base/Substack/2026-07/2026-07-08-tmtb-morning-wrap.md
发言人：KeyBanc 卖方；M-sci 数据（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L234-L245 可信度：仅作局部线索
财报前 KeyBanc 下调 FY27 Cloud 估计（迁移和有机增长假设都下调），预期保守指引是 clearing event；M-sci 称 6 月趋势偏软但新 logo 环比小幅改善。反映财报前卖方对有机增长偏谨慎。
问题：事后看 F4Q 实际大超，说明财报前卖方偏悲观
C 2026-07-08 · GuidePoint · 前 LinearB 客户方案经理：Atlassian Bitbucket 'dying on the vine'；行业都在转向消费制 2 条
文件：Knowledge Base/GuidePoint/2026-07/2026-07-08-linearb-evaluating-this-ai-powered-developer-productivity-platform.md
发言人：Vincent Lussenburg，前 LinearB Senior Manager of Customer Solutions（明示）
阅读范围：全文检索 Atlassian/Jira/Bitbucket/seat 段落 L205-L240、L320-L330、L440-L462 可信度：仅作局部线索
专家称 Atlassian 在 Bitbucket 上投入很少、逐渐萎缩；DX 被 Atlassian 收购后放缓；开发者工具普遍从按人头转向消费制。与 Jira seats 关系弱。
C 2026-07-10 · TMTB (Substack) · BTIG 渠道调研：年末收尾强劲，预期 FY27 Cloud 指引 23% 低于共识 1 条
文件：Knowledge Base/Substack/2026-07/2026-07-10-tmtb-morning-wrap.md
发言人：BTIG 卖方（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L291-L296 可信度：仅作局部线索
BTIG 合作伙伴调研称 TEAM 年末强劲，但模型预期 FY27 Cloud 指引 23%（共识 24%）、DC -13%、总收入 12%；未谈 seats。
B 2026-07-14 · GuidePoint · Notion 客户（Rakuten France）：1.5 年前完全砍掉 Jira，改用 Notion 做工单，成本减半 3 条
文件：Knowledge Base/GuidePoint/2026-07/2026-07-14-genai-series-notion-a-customer-discusses-this-enterprise-productivity-platform.md
发言人：Sidney Golstein，Rakuten France Lead AI PM（明示）；Rakuten France 约 200 人
阅读范围：全文检索 Jira/Atlassian 段落 L55-L60、L130-L134、L350-L425，及背景 L140-L165 可信度：仅作局部线索
Rakuten France（约 200 人）1.5 年前从 Jira 全面切换到 Notion 做 backlog/工单，并用 Notion AI agent 生成工单，称'用 Notion 替代整个 Atlassian 套件'、成本降一半。是中小客户层面竞品挤占 Jira seats 的一手证据，且 AI/context 是迁移理由之一。
问题：单一小型客户（约 200 人），代表性有限；Atlassian 管理层称流失主要在小客户
A 2026-07-29 · Third Bridge · Atlassian 前 VP：AI 创造工作，seat 不降反增；Service Collection 可到 20 亿美元 ARR 12 条
文件：Knowledge Base/ThirdBridge/2026-07/2026-07-29-atlassian-ai-can-write-the-code-but-who-manages-the-work.md
发言人：Bruce Randall，前 Atlassian VP Product Marketing（AI Service Management/AIOps，负责 Service Collection GTM），现 BMC Helix 营销承包商，此前在 ServiceNow（L12-13, L34, L42, L50 明示）
阅读范围：全文 L1-L172 可信度：需交叉验证
前 Atlassian ITSM 产品营销 VP 明确认为 AI 扩张而非压缩 seat，逻辑是 AI 把客户可完成工作从约 10% 提到 30-40% 并“创造工作”，但其 seat 观察主要来自 ServiceNow 时期经验、没有给出 Atlassian 自身 seat 数据。对 TWC“seat 扩张 >10%”和 cloud 29% 增速由 seat expansion 驱动的说法，专家只能说打包红利会持续一段时间，未能拆分可持续部分；$1m+ 客户翻倍被归因于存量合同合并而非新客户。
问题：专家 seat 观察的核心来源是 ServiceNow 时期（L50），对 Atlassian 只说‘similar in this regard’，没有 Atlassian 客户 seat 数据；‘10%→30-40%’和‘两个应用变四个’都是专家自称示意性数字（L54 just making these numbers up）；专家是 ITSM/Service Collection 背景，对 Jira 开发者 seat 和非技术用户占比未谈；TWC >10% seat 扩张、29% cloud、$1m+ 翻倍等数字均来自分析师引用公开材料，并非专家披露；专家现为 BMC Helix 承包商，离职时间未明示
A 2026-08-06 · S&P 业绩会纪要 · FQ4 FY26 业绩会：云加速到 31%，seat expansion 再次成为超预期驱动；TWC 客户每实例 paid seats 是单品的 4-5 倍；但 FY27 sub-ARR 指引从 23% 降到 18% 19 条
文件：/tmp/claude-0/seats/mgmt/2026-08-06_Earnings_Call.txt
发言人：CEO Mike Cannon-Brookes、CFO James Chuong（明示）及卖方分析师
阅读范围：全文 L1-L462 可信度：可作主要证据
云收入加速到 31%，CFO 再次称迁移不是 4900 万美元超预期的来源，驱动是 cross-sell 和 seat expansion（Jira 和 Confluence 核心产品 seat 扩张强劲，约 2/3 用户是知识工作者）。CEO 称财务、HR、市场人员也在用这些工具做技术。TWC 客户每实例 paid seats 是单品客户的 4-5 倍，推动 ARPU 提升。不利信息：FY27 subscription ARR 指引 18%，低于 FY26 的 23%（含 DX 1pt），分析师称隐含有机净新增 ARR 大致持平。Morgan Stanley 问其中多少是 seat 扩张放缓，CFO 只说审慎和宏观，没正面回答。
问题：知识工作者占比口径：本篇 Jira+Confluence 约 2/3，FQ4 FY25 是 core apps 50% 业务用户，口径不同且没解释；'stronger seat expansion'（L89）没说相对什么时期，CFO 回答 Gregg 时也没给环比；TWC 每实例 4-5 倍 paid seats 可能是选择效应（大客户更会买 TWC），不能直接视为 TWC 带来的增量；MCP '400% with barely a blip' 原意可能是服务成本基本没增加，字面歧义；L321-322 '$14 billion (sic) [ old figure ]' 是转录方标注，CEO 引用了旧 TAM 数字；'service question' 为 Service Collection 的转录错误
C 2026-08-06 · TMTB (Substack) · TEAM F4Q26 first take：Cloud 31%、F1Q Cloud 指引 28.5% 远超买方 22-23% 1 条
文件：Knowledge Base/Substack/2026-08/2026-08-06-tmtb-eod-wrap-net-team-akam-mchp-lyft-twlo-first-takes.md
发言人：TMTB 作者
阅读范围：仅检索 TEAM 段落 L36-L45 可信度：仅作局部线索
财报当晚快评，只给增长和指引数字，无 seat 说法；FY27 收入 +13% 略低于 Street。
A 2026-08-07 · TMTB (Substack) · TEAM F4Q26 点评：$49M Cloud 超预期全部来自 Jira/Confluence seat 扩张+Collections 交叉销售 10 条
文件：Knowledge Base/Substack/2026-08/2026-08-07-tmtb-morning-wrap.md
发言人：TMTB 作者转述业绩会与 callback
阅读范围：仅检索 TEAM 段落 L200-L259 可信度：需交叉验证
F4Q26 Cloud +31%（连续三季加速），作者称 ~$49M Cloud 超预期全部来自核心 seat 扩张和 Collections 交叉销售，迁移持平；TWC 客户每实例付费 seats 是单品客户 4-5 倍；约 2/3 Jira/Confluence 用户为知识工作者，技术和知识工作者 seats 均扩张。但 FY27 Subscription ARR 指引 18% 意味有机净新增 ARR 大致持平，空头质疑'审慎'掩盖 seat 扩张放缓；Rovo 超额未收费、AI 直接变现小。
问题：'4-5x seats per instance' 是横截面（大客户更倾向买 TWC），不代表 TWC 导致 seat 增长——可能选择偏差；seat 扩张仍无绝对数或增速；汇编转述
A 2026-08-11 · S&P 卖方会议纪要 · IR 在 KeyBanc 会议（Q4 后）：Jira/Confluence 付费 seat 强劲扩张，净新增 seats 覆盖开发者和知识工作者；FY27 ARR 指引 18% 13 条
文件：/tmp/claude-0/seats/mgmt/2026-08-11_Company_Conference_Presentation.txt
发言人：Martin Lam，Head of IR（明示）；主持 Jason Celino (KeyBanc)
阅读范围：全文 L1-L306 可信度：可作主要证据
Q4 cloud 超预期驱动与 Q3 一致：TWC 升级/Service Collection 交叉销售 + Jira/Confluence 强劲 seat 扩张，且净新增 seats 同时来自开发者和知识工作者（未量化）。分析师追问开发者/知识工作者招聘是否是短期现象、离岸外包+AI 的风险，IR 只给了定性回答。FY27 Subscription ARR 指引 18%（Q4 为 23%），提示减速；Rovo credit 限额今年开始执行，consumption 定价的比重会上升。
问题：seat 扩张仍未量化；L275-276 两个客户群都说成 '$3 million'，按 Citi 会议应为 $3M 增 50%、$5M 增 70%；ARR 23% 中约 1pt 来自 DX 收购（见 Citi 会议），本稿未提
B 2026-08-26 · TMTB (Substack) · RBC 首次覆盖 Outperform：AI 是顺风而非 seat 压缩威胁；AI 编程早期用户 Jira 任务多约 5% 2 条
文件：Knowledge Base/Substack/2026-08/2026-08-26-tmtb-morning-wrap.md
发言人：RBC 卖方（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L291-L297 可信度：需交叉验证
RBC 认为 AI 增加软件创造和协调复杂度，使 Jira/Confluence 更有价值，可维持 20%+ subscription ARR 增长；引用 AI 编程工具早期用户 Jira 任务 +~5%、MAU 更高、seat 扩张更快。
问题：~5% 的口径（对照组、来源）未说明，可能来自公司投资者日材料；5% 任务增量幅度不大
C 2026-09-03 · 专家电话 (音频转写) · 微软战略客户 CTO：Atlassian 在 seat 不增时靠升级版本获益；Rovo 数据转述 5 条
文件：Knowledge Base/2026-09-03/9月3日 微软Chris(audio).md
发言人：Chris，微软面向战略客户的首席技术官，服务 1 家大型私人跨国集团（L9-L11 明示）；自称并非 Atlassian 专家（L1188）
阅读范围：仅检索关键词段落：全文 1283 行，检索 Atlassian/Jira/Confluence/Rovo/席位/seat/ServiceNow/按用户，精读 L645-L780（中文纪要第 11 节）、L1068-L1110、L1168-L1214（中文转译）及 L1284 英文原文中对应 seat 片段；其余（微软 AI 支出、算力、ServiceNow/Salesforce）未精读 可信度：仅作局部线索
分析师直接问‘seat 不增时 Atlassian 如何受益’，专家没有回答 seat 是否在增，而是转向‘推动客户从基础版升级到高级版/企业版’和项目/工程数据价值，相当于默认 seat 不增的前提。关于 Atlassian 把 seat 卖进财务/HR/营销，专家未给证据，只转述 Rovo 公开数据并称跨部门能力‘待验证’。
问题：专家自称非 Atlassian 专家（L1188），观点为外部推测；Rovo 数字应来自 Atlassian 公开材料，不是独立证据；长文件仅检索关键词段落，非全文精读
X 2026-09-03 · TMTB (Substack) · 未提及 Atlassian/TEAM/Jira 0 条
文件：Knowledge Base/Substack/2026-09/2026-09-03-tmtb-morning-wrap.md
发言人：—
阅读范围：关键词检索 Atlassian/Jira/TEAM/seat 无命中 可信度：仅作局部线索
无相关内容。
B 2026-09-08 · GuidePoint · Co-operators 首席 AI 官：Factory 直接接入 Jira/Confluence 是优势；长期用 coding agent 减少工程师人数 3 条
文件：Knowledge Base/GuidePoint/2026-09/2026-09-08-factory-ai-customer-insights-on-droid-coding-agent-capabilities-and-competition.md
发言人：Julian Lustig，Co-operators VP Digital & Data Platforms / Chief AI Officer（明示）
阅读范围：全文检索 Jira/Confluence/headcount 段落 L75-L80、L278-L290、L314-L328、L376-L384 可信度：仅作局部线索
大型保险企业客户：选择 Factory 的优势之一是直接接入 Jira 和 Confluence（Devin 自带工单系统导致两套工单），说明 coding agent 工作流仍以 Jira 为工单中心；但未来 24 个月释放开发产能，更长期预期减少工程师/架构师/科学家人数——直接对应 dev seat 压缩风险。
问题：减员是长期预期，非当前事实
A 2026-09-09 · FactSet 转录 · CFO 在 Citi 会议：Q4 超预期来自 Jira/Confluence/JSM 独立 seat 增长，Jira 约 $7/seat，增长主要来自 seat 数量而非标价 13 条
文件：Knowledge Base/FactSet/2026-09/2026-09-09-near-real-time-transcript-atlassian-corp-team-us-citi-global-tmt-conference-9-september-2026-3-55-pm-et.md
发言人：James Chuong，CFO（明示）；主持 Fatima Boolani (Citi)。转录未区分发言人，按语义推断
阅读范围：全文 L1-L297 可信度：需交叉验证
CFO 称 Q4 超预期主要来自 Jira、Confluence、JSM 的独立 seat 增长，并称看到开发者和非开发者 seats 都在增加；回应提价问题时说主要增量不是标价，而是 seat 数增加、版本升级和单品→collection。Jira 约 $7/seat 被当作提价空间的依据。agent 不按 seat 收费：简单 Teamwork Graph 查询免费，复杂任务消耗 credits，12 月开始执行限额。ARR 23% 中约 1pt 来自 DX 收购，FY27 指引减速被解释为审慎。
问题：近实时转录质量差：'See growth'/'suite growth'/'C-PACE growth' 应为 seat growth；'Non-purpose' 应为非开发者；'Detaxe' 应为 DX；$7 per seat 所在句 L170-171 断句，缺少单位（per user per month?）和完整原话，且与 6/10 IR 说的 $8 不一致；L154-155 'north of 120% year over year growth ... Cloud MRR' 可能是 cloud NRR >120% 的转录错误；转录未标发言人，问答边界需推断
B 2026-09-14 · TMTB (Substack) · Mizuho：最新 Cloud 提价约 7-10% 标价上涨，12/3 起约 10% Cloud 客户或面临用量收费 2 条
文件：Knowledge Base/Substack/2026-09/2026-09-14-tmtb-morning-wrap.md
发言人：Mizuho 卖方（TMTB 转述）
阅读范围：仅检索 TEAM 段落 L321-L327 可信度：需交叉验证
Mizuho 称 TEAM 最新 Cloud 提价意味约 7-10% 标价上涨（略高于去年），扩展的用量计费 12 月 3 日开始，约 10% Cloud 客户可能产生额外费用，并推动更高层级 TWC 采用；PT 升至 $215。
问题：提价是 ARPU 驱动但也可能是一次性/周期性贡献（E05）
X 2026-09-16 · TMTB (Substack) · 'TEAM (Jira)' 仅一张图片 0 条
文件：Knowledge Base/Substack/2026-09/2026-09-16-tmtb-eod-wrap.md
发言人：TMTB
阅读范围：仅检索 TEAM 段落 L114-L117 可信度：仅作局部线索
仅有 'TEAM (Jira):' 加一张图片，无文字内容。
问题：图片内容不可读
B 2026-09-23 · Funda · 美国头部电信运营商战略高管：开发者人数 0-5% 增长、产出 15-30%；Jira 应按人+agent 工作流而非开发者 seat 建模 6 条
文件：Knowledge Base/Funda/20260923 企业 AI 支出 Expert — 平台整合与工作负载带动效应.md
发言人：Strategy Executive，美国头部电信运营商（前任职，L14 明示）
阅读范围：全文 L1-L390 浏览，精读 Atlassian/开发工具段落 L32-L43、L303-L335 及 SOR/外围 SaaS 精简段 L337-L391 可信度：需交叉验证
大型企业侧观察：AI 没有让研发人数同比例缩减，但工程团队基本持平（低个位数增长），开发者 headcount 未来 1-2 年仅 0-5% 增长而产出 15-30%；因此纯按人头 seat 的开发工具承压。专家认为 Atlassian 价值不会下降，但应从‘人类 seat’转向‘seat + agent + 用量’定价，即 seat 增长不是驱动，工作流规模才是。未给出该公司 Jira/Confluence seat 实际增减数字。
问题：未给出该公司 Jira/Confluence 实际 seat 增减或支出数据；‘外围 ticketing/knowledge search 精简’是否包括 Jira/Confluence 未说明；专家是电信运营商战略岗，非 IT 采购一线，数字为方向性估计
X 2026-09-24 · TMTB (Substack) · AI token 价格下降背景下 Atlassian 提醒自托管模型有成本 0 条
文件：Knowledge Base/Substack/2026-09/2026-09-24-tmtb-morning-wrap.md
发言人：The Information 报道（TMTB 转引）
阅读范围：仅检索 Atlassian 段落 L183 可信度：仅作局部线索
只提到 Atlassian 提醒自托管模型有服务与人才成本，与 seats 无关。
C 2026-09-28 · TMTB (Substack) · Cannon-Brookes Decoder 访谈标题：SaaSpocalypse 是错的，MCP 用户扩大 Atlassian 支出速度快两倍 1 条
文件：Knowledge Base/Substack/2026-09/2026-09-28-tmtb-podcast-intel-instinct-founder-on-iltb-nvidia-nvda-s-jensen-altassian-team-s-ceo-cloudflare-net-s-c.md
发言人：TMTB 目录标题
阅读范围：全文 L1-L95（正文仅为图片链接，只有标题文字） 可信度：仅作局部线索
只有访谈标题：MCP 用户扩大 Atlassian 支出速度是两倍；正文为图片，无可读内容。
问题：仅标题，正文是图片
B 2026-10-08 · TMTB (Substack) · In Practise 前 Atlassian 全球企业销售负责人：SMB 裁员导致 seat 下滑约 20% 2 条
文件：Knowledge Base/Substack/2026-10/2026-10-08-tmtb-eod-wrap-the-great-gross-to-net-panic-of-2026.md
发言人：前 Atlassian Global Head of Enterprise Sales（In Practise 推文引述）
阅读范围：仅检索 TEAM 段落 L200-L228 可信度：需交叉验证
转引 In Practise 访谈：SMB 客户以前从 50 人涨到 60 人带来 20% 用户增长，现在裁员从 50 降到 40，seat 下降约 20%。这是对 SMB seat 下滑的直接一手说法，与 thesis 相反；同日 TEAM 作为 seat 型软件随板块轧空 +4%。
问题：50→40 是举例而非统计；未说明 SMB 占比和时间段；仅推文截取，原访谈未读
B 2026-10-12 · TMTB (Substack) · Cannon-Brookes 20VC：Jira 和 TWC 刚经历最好的增长年，用户规模加速；agent 不会取代界面 4 条
文件：Knowledge Base/Substack/2026-10/2026-10-12-tmtb-podcast-intel-zuckerberg-and-wang-on-metas-meta-100b-muse-wonder-bet-wang-on-building-muse-nvidias-n.md
发言人：Atlassian CEO Mike Cannon-Brookes（摘要转述）
阅读范围：仅检索 Atlassian 段落 L40-L42、L371-L442 可信度：需交叉验证
CEO 称 Jira 和 Teamwork Collection 刚经历最好的增长年、用户规模加速；agent 只会替代'平庸'功能任务，协作重的企业工作流仍需界面；Confluence ARR 超 $1.5B、增长 25%+；同时承认企业对 AI 预算越来越价格敏感。
问题：'accelerating at massive scale in users' 是 CEO 定性，无 seat 数字；用户≠付费 seat
X 2026-10-12 · Third Bridge · Dropbox Dash 访谈：仅泛泛提及 Atlassian 推出 agent 0 条
文件：Knowledge Base/ThirdBridge/2026-10/2026-10-12-dropbox-dash-the-next-phase-of-ai-powered-knowledge-work.md
发言人：专家（MS）
阅读范围：全文检索 Atlassian 段落 L87、L131 可信度：仅作局部线索
只把 Atlassian 列为在推 agent 的公司之一，与 seats 无关。