2026.09.22 | Substack | Markets | TMTB Podcast Intel: Benioff
原标题：TMTB Podcast Intel: Salesforce’s (CRM) Benioff on the SaaSpocalypse, Slack and Lab IPOs, Rogo’s Gabe Stengel on AI…
TMTB Podcast Intel is produced by Ethmos, research agents that listen to every episode in TMTB’s coverage universe and write the briefs below.
Today’s Essential Episodes:
- Salesforce CEO Marc Benioff on the SaaSpocalypse, Slack in the Age of AI, Hugging Face and the Lab IPOs
Sources with Alex Heath, and Marc Benioff, co-founder and CEO of Salesforce Benioff after Dreamforce: where he says the software sell-off stands, why Slack matters more in the age of agents, his role in the Hugging Face acquisition, the wealth the Anthropic and OpenAI IPOs are creating, and what the industry should take from the social media era.
- Rogo CEO Gabe Stengel on Building Investing Superintelligence and the Last-Mile Harness
Invest Like the Best with Patrick O’Shaughnessy, and Gabe Stengel, co-founder and CEO of Rogo Why he says models now do most of what a junior banker does, so the harness, compliance plumbing and distribution become the moat; deal-makers over public-equity investors as the first customers; the move from seat-based to outcome-based pricing; and 40 VC passes before the Series A.
- Vinod Khosla on the Personal-Agent Market: Trust, Task Completion and Who Wins the Agent Fight
TBPN with John Coogan and Jordi Hays, and Vinod Khosla, founder of Khosla Ventures (the Khosla segment, recorded after Amazon blocked Meta’s Muse) Why he sees a broad, non-winner-take-all agent market, the fight between user-aligned agents and platforms defending their own agendas after Amazon blocked Muse, whether consumers will pay, and why trust, task completion and secure payments, not data lock-in, decide who wins.
- TypeSafe AI’s Diogo Almeida on Jev, the ‘System One’ Model Built for Production, Not God
Latent Space with Shawn Wang (swyx), and Diogo Almeida, CEO of TypeSafe AI and InstructGPT co-author A non-autoregressive model built for code, not chat, after a 40M-view launch: why he says RLHF causes mode collapse and unreliability, why Jev skips public benchmarks, refusals and fine-tuning, and the intelligence-per-dollar case against frontier chatbots.
- Naveen Rao on the Energy Wall and a Chip That Merges Compute and Memory
All-In: a talk by Naveen Rao, CEO of Unconventional AI (formerly of Databricks, MosaicML and Nervana), followed by Q&A with Chamath Palihapitiya Why he says AI hits an energy wall before a compute wall, the ‘dynamical computer’ that collapses the memory-compute divide, a prototype generating images at roughly 500 nanojoules each, the 1,000x efficiency target, and a data-center product he puts within two years.
Read the Brief on Ethmos | Listen | Watch
Read the Brief on Ethmos | Listen
Read the Brief on Ethmos | Listen | Watch
Read the Brief on Ethmos | Listen
Read the Brief on Ethmos | Listen | Watch
TMTB Podcast Intel is Powered by Ethmos | Podcast Research Agents
Try Ethmos Free