The All-In Podcast's latest episode, published May 11, 2026, runs 1 hour 11 minutes and covers four major topics: the SpaceX-Anthropic compute deal and Elon Musk's cloud ambitions, the question of whether Anthropic is forming a structural AI monopoly, the Washington panic over an FDA-style AI regulatory body, and how the besties are thinking about trading the AI market cycle. Hosts are Chamath Palihapitiya, Jason Calacanis, David Sacks, and David Friedberg.
Full episode via All-In Podcast on YouTube
All-In Podcast Recap | SpaceX-Anthropic Deal and Elon Web Services (4:38)
The main segment opens with the SpaceX-Anthropic compute agreement, in which SpaceX will supply GPU infrastructure to Anthropic for model training and inference. The panel frames this as the first concrete step toward what Jason Calacanis calls "Elon Web Services," positioning Musk's infrastructure assets, including SpaceX's compute build-out, Starlink connectivity, and xAI's own model stack, as a vertically integrated cloud platform that could challenge Amazon Web Services and Microsoft Azure for AI workloads.
The discussion also covers Anthropic's growth trajectory, which the panel describes as unprecedented in enterprise software: faster revenue ramp than Salesforce, faster than Workday, and accelerating rather than slowing. Chamath argues this is the most important data point in tech right now. The SpaceX IPO valuation question runs through the segment, with the besties debating how a cloud infrastructure business changes the multiple.
Is Anthropic the Next AI Monopoly? (26:48)
The second segment picks up the Anthropic thread and asks a harder question: is the revenue growth a sign of permanent structural dominance, or is it a moment in time that OpenAI and Google can disrupt? David Friedberg takes the skeptical view, pointing to the history of technology markets where first-mover advantages in infrastructure eroded quickly once hyperscalers commoditized the underlying capability. Sacks and Chamath are more bullish on Anthropic's moat, citing Claude's enterprise adoption in legal, finance, and software development, sectors where switching costs are high once workflows are built.
FDA for AI Panic and White House AI Safety Thinking (35:21)
The panel addresses a proposal circulating in Washington to require pre-release federal approval for frontier AI models, modeled on the FDA's drug approval process. All four besties oppose the approach, with David Sacks providing the most detailed window into administration thinking: the White House favors voluntary safety commitments from labs and an America-first competitiveness frame over creating a new federal agency that could slow US development while China accelerates. The segment references a New York Times report from May 4, 2026, on the Trump administration's AI safety policy direction, and the White House national policy framework document published in March 2026. The panel also spends time on the positive flip side: specific cases where AI is already delivering measurable outcomes in charitable giving platforms, early cancer detection, and personalized education tools.
Trading the AI Boom | State of the Economy (1:00:04)
The closing segment covers market positioning. Chamath lays out his current thesis: the market is not yet pricing in a winner-take-all outcome in AI infrastructure, and the most asymmetric trade is in companies supplying the physical layer, power, cooling, and networking, rather than the model layer itself. Friedberg is more cautious on valuations across the board, noting that the AI trade has already re-rated significantly from 2024 lows and that incremental buyers are taking on more duration risk. The full episode is available on Spotify and YouTube. For more creator and podcast coverage, see the oWire Creator hub.