All-In PodcastWhy Congress voted 427-1 to release the Epstein Files
An overwhelming 427-1 House vote forced the Epstein Files into the open; one host admitted being in Epstein's black book and suspects he was a spy.
CHAPTERS
- 0:00 – 4:20
Vegas Setup, F1, and Venetian Poker Plans
The hosts open from a studio at The Venetian in Las Vegas, joking about being together in person for F1 and high-stakes poker. They plug The Venetian’s new poker room, tease private games with pros like Phil Hellmuth and Jason Koon, and banter about their home game dynamics.
- •All four besties are together in Las Vegas for Formula 1.
- •The Venetian is hosting them with VIP suites and a high-tech poker room.
- •They tease an afternoon of secret poker games featuring top pros and friends.
- •Light banter sets a casual tone before pivoting to heavier topics.
- 4:20 – 14:00
Epstein Files: Politics, Precedent, and Public Pressure
They unpack the near-unanimous Congressional vote to release the Epstein files and Trump’s ‘give them everything’ reversal. The conversation explores why damaging Trump material likely doesn’t exist, why Democrats may be more exposed, and whether releasing these files sets a precedent for opening other sensitive investigations.
- •House passes Epstein Files Transparency Act 427–1; Senate unanimous consent; Trump signs.
- •Only dissenter Clay Higgins warns about hurting innocents and breaking 250 years of criminal procedure.
- •Chamath argues absence of Trump bombshells suggests more risk to Democratic elites and operatives.
- •They contrast legitimate victim protection with the need to honor public demands for transparency.
- •Discussion of whether Epstein is a singular case or a precedent for future investigative file releases.
- 14:00 – 37:00
Epstein’s Networks: TED, Money, and Intelligence-Angle Speculation
Jason recounts meeting Epstein at TED ‘billionaires’ dinners and being in his infamous black book, describing how the tech and science community initially framed Epstein’s Florida conviction as a misunderstanding. The group analyzes his role as a donor to scientists and tax adviser to billionaires, questioning the true source and purpose of his wealth and relationships, and raising the possibility he served as an intelligence asset.
- •Jason met Epstein several times at TED billionaires’ dinners in the late 1990s/2000s.
- •Epstein was pitched within that circle as a victim of a ‘set up’ in Florida, not a serial abuser.
- •He gave money to MIT-affiliated scientists and advised tech billionaires on tax strategies.
- •The Leon Black $168M ‘tax advice’ payment appears outsized versus typical estate-planning fees.
- •Jason predicts files will reveal intelligence-agency entanglements and embarrassing revelations across parties and academia.
- •They stress not equating everyone who took Epstein’s money or met him with criminal complicity.
- 37:00 – 43:40
Dinner, Tether, and the Stablecoin Mega-Business
Shifting from Epstein to crypto, Chamath describes a dinner with Tether CEO Paolo Ardoino and outlines why he sees Tether as an astonishingly powerful business. They walk through how dollar-backed stablecoins work, their role as a hedge against inflation in emerging markets, and the massive interest spread captured by issuers.
- •Chamath praises Paolo Ardoino and calls Tether an ‘incredible business’.
- •About 500M people use dollar-backed stablecoins, heavily in Africa, Central America, and Asia.
- •Users swap devaluing local currency for USDT; Tether invests backing dollars in US Treasuries.
- •Tether reportedly holds ~ $135B in Treasuries plus Bitcoin, gold, and land, earning billions in interest.
- •End users typically earn no yield; they value stability over interest income.
- •Jason notes he was previously highly critical of Tether (lack of audits, bans) but credits recent cleanup and moves toward audits.
- 43:40 – 53:50
Regulating Stablecoins: Banks, Yield, and the Coming Competition
They dive into the policy fight over who can capture stablecoin yield and what U.S. regulation should look like. The conversation highlights banks’ efforts to block interest-bearing stablecoins, crypto firms’ ‘rewards’ workaround, and David Sacks’s role in designing a more constructive U.S. framework after a decade of anti‑crypto policy.
- •Clarity Bill in Congress will define U.S. stablecoin market structure.
- •Banks fought to prevent stablecoin issuers from paying interest directly to consumers to protect net interest margins.
- •Crypto firms currently use ‘rewards’ programs as a quasi-yield workaround.
- •Tether must unwind bans in New York, Canada, and other jurisdictions to enter the U.S. fully.
- •Chamath invites Jason to El Salvador for a Tether conference and an interview with President Bukele.
- •They expect competition from Stripe, Visa, Circle, and others to erode Tether’s extraordinary margins over time, especially as rates fall.
- 53:50 – 1:07:00
Nvidia, Burry’s Short, and GAAP Depreciation Explained
The hosts pivot to Nvidia’s blowout earnings and Michael Burry’s short thesis, which centers on alleged under-depreciation of GPUs and overstated tech profits. Friedberg walks through GAAP Accounting Standard 360, arguing that Burry conflates technological obsolescence with useful life and ignores that all relevant cash flows are visible to investors.
- •Nvidia reports massive year-over-year and quarter-over-quarter growth; GPUs are sold out.
- •Burry claims big tech is ‘cooking the books’ by using overly long useful lives for data center GPUs.
- •Friedberg explains that under GAAP, depreciation must track actual revenue-generating use, not just tech advancement.
- •Difference between 3-year vs 6-year depreciation would shave maybe ~10–12% off Google’s profit, not collapse it.
- •Cash flow statements already disclose all capex; investors can and do value firms on free cash flow, not just EBITDA.
- •Chamath criticizes Burry for not digging into how AI systems monetize ‘output tokens’ and for lacking technical literacy about model economics.
- 1:07:00 – 1:23:00
Google Gemini 3, TPUs, and the Fragmenting AI Stack
They analyze Google’s Gemini 3 release, which is believed to be trained exclusively on TPUs, and the resurgence of Google in AI benchmarks and chat share. The hosts project a future where multiple specialized chips and models serve different domains, assess risk to Nvidia from hyperscaler silicon, and outline why OpenAI may be structurally disadvantaged.
- •Gemini 3 reportedly trained on Google TPUs; Google now leads many LLM benchmarks.
- •Google’s chat share climbed from ~8% to ~16%, undercutting the ‘ChatGPT kills search’ narrative.
- •Chamath expects a fragmented decode layer: Google TPUs, xAI’s Grok chips, Microsoft, Amazon, and Meta custom silicon.
- •Friedberg predicts Huawei will emerge around 2026 with competitive, AI-designed chips built in Chinese fabs—potentially a major Nvidia risk.
- •Jason argues OpenAI is losing enterprise trust because it competes at the application layer (e.g., Sora, coding tools), pushing startups to Anthropic and open source.
- •They frame a notional trade: structurally bullish Google, Anthropic, and Grok; cautious or ‘short’ on OpenAI’s long-term dominance.
- 1:23:00 – 1:47:00
VC vs. Operating: Power Laws, LPs, and Friedberg’s Oppenheimer Moment
The conversation turns introspective as they compare returns from managing outside capital versus investing personal money. Chamath explains why his solo investing has higher dispersion but more upside, while Friedberg describes his journey from venture studio to going all-in as CEO of Ohalo after questioning his life’s impact post–Oppenheimer.
- •Chamath’s personal investing yields higher returns but also higher volatility versus fund management.
- •As a GP managing LP money, he optimized for ‘never lose money’ and quick return of capital to institutions like Mayo Clinic.
- •As an individual, he tolerates large drawdowns (e.g., a $400M loss in Relativity Space) in exchange for bigger upside.
- •Friedberg recounts earlier ventures (MetroMile, Eatsa) that misled him into thinking he could scale as chairman-only.
- •Years of board frustration convinced him he needed to operate again; Ohalo’s breakthrough and seeing Oppenheimer crystallized that he wasn’t fulfilling his potential just as an investor.
- •LPs welcomed his move to become Ohalo’s CEO; his venture studio is evolving into a holding company primarily owning Ohalo.
- 1:47:00
High-Stakes Poker with Alan Keating: Fear, Soul Reads, and Deep Water
In the Vegas segment’s climax, they bring on high-stakes pro Alan Keating, known for hyper-aggressive, unorthodox play on streamed games. They dissect his now-famous call versus Doug Polk with 4‑2 offsuit, explore how he reads fear and physical tells, and discuss why he deliberately drags opponents into ‘deep water’ where preparation fails and psychology dominates.
- •Keating, a former Big Game runner and seed investor in Polymarket, joins the table.
- •Chamath defends him as an ‘exceptional’ live player with Hellmuth-like soul reads, not just a maniac.
- •Keating explains that he cares less about solvers and more about what’s happening with a person in the moment.
- •In the Doug Polk hand, he relied on prior bet-sizing and cadence patterns to conclude Doug was weak, calling a huge river bet with 4‑2.
- •He emphasizes ‘navigating fear’ better than others: when people are scared, they telegraph and make bad decisions.
- •Keating admits he likes bets where failure would put him ‘in a little bit of trouble’—pressure motivates him.
- •They tease more poker content from The Venetian with pros like Jason Koon and Phil Hellmuth.
