Industry veterans, degenerate gamblers and besties Chamath Palihapitiya, Jason Calacanis, David Sacks, and David Friedberg cover all things economic, tech, political, social, and poker.
When a partnership with Rolls-Royce collapsed and everyone said Boom Supersonic was finished, the forced pivot to building their own engine turned into a second, unplanned, faster-growing business.
A squirrel's brain runs flawless real-time motor control on 8 nanowatts. Naveen Rao's new chip company built a working prototype in five months by asking why computers move so much more data than brains do.
When investors stopped listening, AppLovin's CEO stopped pitching them and started buying back the company's own stock instead, turning a 92% crash into a run to $250 billion.
Investor Bill Gurley studies five major disasters, from a condo collapse to the 737 Max, to extract a repeatable pattern for how institutions investigate their own failures, then argues COVID broke that pattern entirely.
NASA's administrator explains why the agency's real problem was never budget, it was trying to make every district and partner happy, and how he's retaining talent by pivoting the agency toward work private industry structurally can't do.
A 24-year-old with no journalism training built one of the most-viewed investigative accounts in the country by knocking on doors, and California responded by passing a law specifically to stop him.
Altimeter's Brad Gerstner lays out the exact monthly revenue number he's watching to decide if AI infrastructure spending is sustainable, and why enterprises increasing AI spend 17x in 18 months isn't the demand-side risk everyone's worried about.
Meta's president walks through the specific compact (pay for your own power, use less water than the farmland it replaced) that turned one Louisiana parish's data center opposition into teachers getting $50,000 bonus checks.
SpaceX's president explains the "clear the chaff" management model that's kept engineers 24 years deep, and Elon Musk calls in to propose a specific, low-trust-required peer-review mechanism for AI safety modeled on how the movie industry avoided government censorship.
The Vice President's "if you built Frankenstein, don't ask for a global governance body, give defenders the tools to fight it" argument, plus how a $9 billion no-questions-asked federal wire transfer became a case study in eligibility-verification failure.
Microsoft's CEO explains why he thinks the "reward hacking" behind the Hugging Face incident is a solvable engineering problem, not a mystical one, and why he's deliberately not chasing a frontier model.
Nvidia's CEO lays out why he thinks AI doom predictions have a consistent track record of being wrong, explains his "go as high as we need to, as low as possible" platform strategy, and gets a surprise live call from President Trump mid-interview.
The besties dig into whether AI doom narratives are organic fear or a moat-building play, whether OpenAI's math breakthrough quietly learned from users' own proprietary reasoning, and how Nike lost $200 billion by drifting from the product and channels that made it dominant.
OpenAI ships GPT-6 (Astra) and the hosts debate what a two-tier AI market means for anyone building on it, whether this is 1998 or 1999 for valuations, how to read the Hugging Face agent "escape" that spooked the public, and why New York banning AI in schools is really a fight over adaptive learning. A wide-ranging episode with a few genuinely useful lessons for product builders.
Months after the "SaaS is dead" panic, Salesforce ripped 20% in a day. The besties unpack why the systems of record survive the agent era, why Benioff is willingly letting Claude become the front end of his own product, and what that means for how PMs should build.
Mathematician Eric Weinstein argues American science has quietly become a "precariat," a workforce too afraid of losing tenure, grants, and reputation to make real contrarian bets, and lays out a case for funding research the way a hedge fund manages a portfolio instead.
The White House's science and technology policy chief argues the real constraint on American research isn't funding, NIH's budget more than tripled since 1998 without a proportional rise in breakthroughs, but the review process itself, and lays out the incentive redesign built to fix it.
Anthropic's Dario Amodei broke his silence to defend himself against regulatory capture accusations, and the All-In hosts use it to dissect a harder question for anyone building AI products, why doomer messaging keeps backfiring into the exact backlash it was meant to prevent.
Flock Safety CEO Garrett Langley explains why the company built an internal tool that got nine Georgia police officers fired for abusing their own product, and why he's deliberately slow walking AI features despite pressure to ship faster.
Everyone hears the same episodes.
Few can do what they describe.
