← 返回

2026.10.07 | TB | Datacenter | AI Datacenter Expert: Hut 8 & Jacobs

原标题：AI Data Centre Industry – Hut 8 & Jacobs – Former Senior Executive, EPC at Hut 8 Corp

#datacenter

Third Bridge · 2026-10-07

Client-led • 18 Sep 2026

Companies: J US, HUT US, IREN US

Expert Former Senior Executive, EPC at Hut 8 Corp

Agenda

Introduction

Client [00:02:13]: I've been spending some time digging into the market around data centres, and in particular, Hut 8. We'd love to cover Hut 8 as an organisation, as well as its leadership team and ability to execute and corporate maturity. We want to understand if this is a company that can ultimately succeed, especially around its new strategy where it's taking less of an active, direct role in running data centres. We're thinking through conversion opportunities in the existing Bitcoin mining sites, and the opportunities and pitfalls there. Have you been following the recent deals and strategy?

Expert [00:07:21]: I'm somewhat familiar. I'm following the industry pretty well, but I'm not looking at their earnings reports or any of that stuff.

Client [00:07:32]: Hut 8 hired a bunch of folks from the energy industry, such as NextEra, as an energy banker. The strategy is deal-making and project management. The company is signing deals where it identifies and selects a site and negotiates a power agreement and the option from the original developer or someone who acquired the site to begin with. It has relationships with Anthropic, Google and Nvidia. Hut 8 brings these companies together and gets a lease and power agreement done for a site. It's using Jacobs as its EPC [engineering, procurement and construction] company and handing off all the construction work to Jacobs. I think Vertiv is supplying a bunch of the equipment.

The data centre gets completely passed off to the tenant. Hut 8 will be responsible for the roof and legally own the building, but it won't do anything to run or equip it. The company has done a couple of these deals and wants to do more. It might start doing some behind-the-meter stuff, but the idea would be to work with different parties to build, implement and manage that. It's not really running its own data centres anymore. That's the context in which I'm trying to think through Hut 8's corporate capabilities, leadership team and ability to execute.

Expert [00:09:55]: Thanks for sharing that exactly tracks my, I guess, perspective of what they're doing is. That's perfect. Yes.

Questions

00:10:02] How much confidence do we have in Hut 8's ability to execute on the deals it has signed, based on its actual responsibilities? How much confidence do we have in its ability to source more deals and be a winner in the market, given the competition level and the fact that there are plenty of other folks trying to do this, too? Is the company likely to be able to pull more stuff together and execute on that? I'd love to get your perspective on Hut 8's organisation, leadership team, the way it did business, what it was good at and what it wasn't, as well as its ability to succeed in this new model.

Expert [00:11:33]: I think that they are positioned exceptionally well. I think what they shared recently is 8GW or 10GW, and that just means what they've pushed off a high confidence or passed through a high confidence behind that gating factor is probably a bunch more. I think you mentioned NextEra, huge, deep energy roots, these guys are just rock solid. I would say that is how I feel so confident as these guys have relationships. They know where the energy is. They know how to really build the PPA or even go through the due diligence to make sure that's even 5-10 years ahead of not anybody else, but around most. That's why everyone is coming to them. Comparatively, an Iren also has a very similar model. That's somebody else I would consider in your guys' analysis, they have a very similar model, also have a massive gigawatt pipeline of energy. That to me is the biggest moat that everybody else right now is scrambling for. They just can't find the energy or the capacity.

These guys essentially have that, say, I guess, funnel of energy coming right over the next 5-10 years, whatever it is. To me, that is the biggest win, and this is where a lot of these Bitcoin miners are finding success because they've already locked up these multi-gigawatt-year deals ahead of time. That's why you're seeing this conversion happen. Besides the team, which, again, I think is very exceptional, I think Asher and Mike Ho, although they are young, incredibly intelligent people, have already proven to show that they can make a couple of these large deals work. I think the Louisiana one, and there's one in South Texas, I believe, that just got announced or got commitment. That's already starting to prove, hey, they're able to convert into a deal. Maybe the lack part is the actual execution, but I would question how much of that is their responsibility if you've got a Jacobs and a Vertiv and you've essentially passed off some of that to somebody else. I think the execution across the entire industry is in question right now. There's just so much build happening. It's not just to one team. It's just how can the supply chain take this much copper or this much steel. It's just insane to really think about.

Then just the headcount that's needed for all these builds across the world is also, I can just tell you, people are desperately searching for headcount, flying people all across the world just to get these builds moving. It's just a question of do they have the personnel and the actual org within the execution team to help drive success through their contractors? Before they were a little bit more like a GC to Bitcoin mining, but now they're somewhat stepping back a little bit and allowing the GCs technically to do it. Again, I'm not in the day-to-day, so I don't know where that level is. I know they used to just do whatever needed to happen to get the Bitcoin mining sites online. Now they've got a legit GC in there, but they're not going to let them just go hop on and start work machinery. There are a little bit more rules involved, not so much cowboying it.

00:15:34] Do you have a view on Jacobs as the GC [general contractor]? How do you feel about Jacobs taking on the build?

Expert [00:15:56]: I don't know what they are exactly, but I would say top five, top three GCs in data centres. They picked the right partners. I think they're set up for success. The question is, again, headcount to do is ask, "Okay, everything sounds great on pay for the contract." Jacobs for sure, top, I would say, three GCs for data centre build. They've been doing a long time. They've got a lot of success. The fact that Asher chose them, I think, is just it goes both ways. Great choice by Asher, who made the decision and just also proves that Jacobs is right for the job.

00:17:12] I've heard that there can be some squirrelliness about these contractors in the industry, and you have to really keep them in line. What is your confidence level in Hut 8's team to supervise Jacobs and keep an eye out?

Expert [00:17:40]: I don't necessarily know the whole team, but great guys, super intelligent, young men being at the forefront of Bitcoin and now AI, the right guys for the job. They just don't have the 15-20 years of experience of somebody who's built data centres for that long. That's the only question I would have is just, yes, a lot of it makes sense, but sometimes it's just levels and there's strategy that's knock-on effects. You know the right questions to ask to make sure to keep these guys in line, not as simple as just every Monday hopping on a call and just having them tell you, yes. Yes, okay, but tell me how yes. How are you on schedule?

It's just, do you have the experience to see through some of the answers and not just relying on somebody who's telling you yes? That all is fine until you make the deadline and it's how do we not see this and you reverse engineering and you see all the signs. We should have been asking all these questions. I think the test will be the squeeze at the end of the site. This is the first large-scale AI they've done.

00:19:13] I don't know if it has, but if Hut 8 has hired a cost-consulting firm to supervise, is that good enough or is that only part of the puzzle?

Expert [00:19:34]: Sometimes the off-take of the client, Fluidstack or Google, for Louisiana, I think, who's doing that deal is they come back and ask for a request for some changes. Then the changes cost money, and then that's how Jacobs might add some additional costs that they didn't take into consideration the first time. Just this is how these things blown out of proportion is through that kind of stuff. It's not necessarily anybody's fault. It's just how some of these deals work. Then (audio distorts) great, but they're not going to be here.

Here's the numbers. Go do something about it. It's already too late. The contract is already written. Yes, again, it's probably better for the next ones. You probably have some lessons learned from the first one. Again, I don't know if the first one is going bad necessarily. Just knowing what I know about the industry is usually there are a lot of lessons learned after doing something at this scale, but they certainly have the (audio distorts) in place. They potentially could just knock out of the park the first time. I don't know if they are using a cost consultant. I would imagine they probably have somebody. I would imagine Jacobs is incentivised to not run over budget and be on time.

00:26:19] Could you talk about Asher? I know he's young. What's his motivation? Is he going to want to build lots of cool stuff? Is he more of a dreamer? Is he all about near-term payback? What's your sense of him as a leader? As this company gets bigger, what is he going to want to do with it?

Expert [00:27:08]: Again, I think he's super intelligent. I think he's proven himself, especially, with this transition from a Bitcoin miner to (audio distorts), whatever it was, USD 200m-some conversion package. He's always asking the right questions just on how the team is going, what we could do to be better. We want to be essentially the company that is going after and converting energy atoms essentially to the best highest profit. Right now, we see Bitcoin mining is a great option. Before the AI, we see AI as a thing that's coming up, and we're open to doing that. In the future, it could be green hydrogen or it could be something else. We're open to do really anything, but we want to just really lock in the energy piece of all this, be an energy-first company and then figure out what the right way is to convert those energy assets into essentially (audio distorts).

00:30:37] Were you surprised that he was able to do amazing deals? Where you not surprised that the team ended up making this work?

Expert [00:30:53]: I was a bit surprised, to be honest. Not in the sense that I just don't believe in them, but just in the sense of like they're coming from a Bitcoin mining company. They've increased the culture, it seems like. Yes, I was a little bit shocked that somebody was willing to fund the project for one and then also that they brought in these larger names, but no one I know now about the demand behind the AI. It's pretty inevitable. I was surprised that some of this, with the boom of AI or even data centres, was able to bring on these kind of deals.

00:32:01] I know there's a ton of demand in the market. Everyone is out scouring for data centre sites. Hut 8 has an advantage here. How do you think about its ability to continue to source sites and deals in this environment where a lot of folks are looking?

Expert [00:32:35]: I think they are well-positioned for sure. I will say that people have now realised the hyperscalers as well, power is the factor here. They've got an explosive next couple of years and everybody else has also realised, oh, my gosh, you got to get power. They get less funnel after that five-year mark, if that makes sense, because now the hyperscalers and everybody else who has a lot of money are now also chasing the same power deals that are coming 8-10 years down the road.

00:33:31] Could you talk about Hut 8's 8GW-plus pipeline?

Expert [00:33:51]: Hut 8 was always well-backed and never broke any promises that I was aware of. I think that helped them originally. That's why I think they had this insane 8GW-plus pipeline right now. Going forward, I think it's a little bit changing that, sure, Hut 8 probably becomes as well as an Iren and these other big players, probably pretty comparable to a hyperscaler. They're going to pay their bills, they're going to take on the project. That plant, it becomes more of a profit game. How much will you really pay us to take the energy and lock it up from us. Where before it was a little bit more of not many people were reaching out saying, "Yes, we'll consume all 400MW or whatever gigawatts of your solar project or whatever it might be." No one was really thinking along those lines other than these large-scale Bitcoin miners, which is why they have the moat and the advantage right now.

00:43:29] How would you think about the Vega site as something that you could or could not turn into an AI data centre?

Expert [00:43:48]: If I remember correctly, it's one large building, plenty of space to essentially convert either throw in Bitcoin ASICs or throw in GPU racks. I think the internals allow for that to happen. I don't know exactly the redundancy built into the mechanical electrical system, but essentially, the design is the same where they have, I think, dry coolers on the outside. You would just need to run probably a couple of additional ones, tap it on to the loop to make it N+1, tier 3 as the off-takers want. They certainly have a substation nearby. I forget, I think it's a wind farm out there, if I'm not mistaken.

Client [00:44:49]: My understanding is wind doesn't work well for AI because you need full availability or five nines of uptime.

Expert [00:45:00]: Yes, something like that. My exact point is, I would question, is there a utility connection there? Great in the sense that you can take all that wind whenever it is spinning, which is renewables. I'm seeing a lot of off-takers prefer that, but you would also have to make sure that that is connected into utility to provide the uptime. You're exactly right. Then the question would be the electrical redundancy, they definitely have one side, the A side, you would need to have a B side as well. Bring in an entire additional electrical pathway. It can come from the same substation, but you'll just need to have, again, another row or another line of copper onto the PDU, back to the switchboard. Just rebuild the entire electrical or to double the entire electrical. Adding generators, they probably have generators now. It really just depends, USD 3.5m-6m per megawatt to convert these Bitcoin mining sites.

Client [00:46:30]: That's not bad.

Expert [00:46:32]: Because a lot of it's already there.

Client [00:46:35]: It's a lot cheaper than a greenfield.

Expert [00:46:36]: Yes, because they already have a lot of sunk in costs. They already thrown in USD 300,000 per megawatt maybe, but yes, they're building like this. I think that's comparable to an Iren. I think Iren basically built USD 4.5m if I'm not mistaken, per megawatt to convert their sites over. I'd say it's probably pretty similar. The next question would just be fibre. Is fibre already on site? Typically, these Bitcoin miners use Starlink because that's all they need.

Client [00:47:02]: The availability of multiple redundant fibre lines from different providers is the biggest concern. Do you know if fibre is available around there, or is that not something you looked at?

Expert [00:47:22]: I do not recall. You probably almost certainly can go into AI and try to find some fibre maps and (talking over each other) call them.

Client [00:47:34]: I've been looking at some of the different data centre sources this week, and I saw some fibre maps.

Expert [00:47:49]: They're not too far from Amarillo. Inevitably, it's got to have fibre somewhat outside of town, I would assume. I just don't know exactly where it is. Is it 100ft away? Probably not. Is it 1-2 miles on the road? Probably. That's not a long run essentially. Yes, I would say the conversion could probably take, I forget what the size of the site, I think 200MW.

00:48:17] What about other Bitcoin data centres? How would you think about the rest of the sites? Is there a really good opportunity to convert them?

Expert [00:48:38]: Essentially, they all can be. It's just a matter of how much it costs. If you have the substation and you have the power there. The question is how much of the infrastructure can you reuse? Really the three pathways you need to confirm is, power, substation, fibre and then all the infrastructure is just an engineering problem at the end of the day, and then a capital problem. I would say they're likely all possible to (audio distorts) I think Salt Creek, which is pretty far out from West Texas. I want to say we did bring fibre over there, but I'm not 100% sure, but yes, that's the 64MW containerised build. Technically, you could throw in containerised GPU builds. Again, the question would be fibre and can you run redundant pathways to the GPUs essentially? Right now, they're all the single pathways coming in, not impossible, just either we take another substation or you would have to cut the project in half from 64MW to 32MW essentially, so you have the power going to each side of the rack. Then I forget the other sites that they have.

Client [00:49:57]: There's King Mountain or something.

Expert [00:50:04]: Yes. King Mountain is very similar. I think it's based off of a wind farm, not too (audio distorts) if I remember right, pretty far, but a very similar conversion as a Salt Creek. It's containerised builds on a bunch of ground in the middle of nowhere, wind farm, and I think a solar farm in the background could just pull out that those containers drop in (audio distorts) rerun some power and some generators and you're probably bringing silicon if you really wanted to.

00:50:34] Has Hut 8 started the process of converting some of its sites at this point?

Expert [00:50:47]: Yes and no. I think they're diversifying. If we ever have a slowdown or anything else, we can quickly convert these over is at least that's how I would do it. They've got whatever it is, 2GW under development now is, let's just focus on those while these sites continue to give us some money or Bitcoin. If for whatever reason we need to bring up some revenue really quickly in one year, we can easily convert one of these over. That's how I would do it. That's probably what they're thinking.

00:51:39] If we think about all the big ex-crypto miners and data centre operators, who do we want to back? How do you feel about Hut 8 succeeding at getting its existing pipeline done and doing stuff on time and executing?

Expert [00:52:19]: I think they're already proven themselves somewhat in what they can essentially own or partially own, they could always have owned the GCP as well, but they hand that over. If those contracts are available or not or if there's any press around, if there's any penalties associated, that's what I don't know. That's where I'd be maybe diving into, but I would say it's not any more unlikely than anybody else. What I mentioned is there is an Iren out there that also has a massive power pipeline, but to my understanding, it has diluted the company.

Client [00:53:17]: You have to believe in the neocloud model. You have to look at the real estate and GPU play.

Expert [00:53:27]: Yes. Exactly. Just a little bit different flavour there, I guess. Yes, that's something I'd maybe compare to is that model vs that model. I know like Marathon now as Mara is getting into the space. I don't know if they've made any large deals yet, but also, I would say, are contender. Just the philosophy of energy is basically the decider for the next 10 years, and they went after and secured as much as they possibly can, and they're willing to shift business models if needed for me is, that's what you want. Laser focus until there is new data to say, hey, we should be focusing on something else, which is a little bit maybe disheartening, but also I think at the end of the day, what they're telling us the company is, which is essentially a power aggregator. Then the off-take can be Bitcoin, AI or green hydrogen or whatever the next awesome thing might be, to me, that's probably a good signal. They're going to offload that conversion responsibility. I think it makes a good (audio distorts) I would certainly take.