All-In PodcastWhy OpenAI needed a federal backstop, not a bailout
Sarah Friar cited a federal backstop in the Wall Street Journal; the actual plan is a buildout, not a bailout, spread across six years and multiple partners.
CHAPTERS
- 0:00 – 9:00
OpenAI’s $1.4 Trillion Spend and Sam Altman’s ‘Feisty’ Interview
The Besties open with Brad Gerstner’s viral BG2 interview with Sam Altman, where a question about OpenAI’s $1.4 trillion in capex commitments sparked market jitters and social media backlash. They unpack Altman’s response, discuss whether his tone was joking or defensive, and explore how such huge numbers fit against OpenAI’s reported revenues.
- •Brad’s ‘softball’ question: how do you justify $1.4T in commitments on ~$13B revenue?
- •Altman pushes back, hints revenue is higher, and jokes about buying Brad’s shares.
- •Brad says the internet over‑indexed on Sam’s tone instead of the substance.
- •The Information reports leaked projections of >$100B revenue for OpenAI and Anthropic.
- •Brad models the 1.4T as 5–6 years of spend, half borne by partners, roughly $150B/year equivalent in later years.
- 9:00 – 16:30
Is There an AI Bubble or Just a Market Repricing?
Chamath and Brad contextualize recent sell‑offs in AI‑adjacent stocks as part of a broader shift to ‘risk off’ rather than a direct indictment of AI fundamentals. They argue investors are grappling with the ROI of enormous capex and how it flows through to future earnings, particularly for the Mag 7 tech giants.
- •Market is digesting massive AI capex and building ROI models under uncertainty.
- •Divergent AI performance among big tech: Google strong on AI; Meta and Apple more mixed.
- •Year‑end dynamics: tax‑loss harvesting and profit‑taking now start in mid‑November.
- •Chamath frames this as a normal risk‑off phase, not Brad/Sam‑driven panic.
- •Brad notes that despite volatility, AI infrastructure equities remain structurally attractive long term.
- 16:30 – 25:00
OpenAI ‘Backstop’ Fury: Bailout Myths and Infrastructure Reality
The hosts dissect OpenAI CFO Sarah Friar’s Wall Street Journal comments about a government ‘backstop’ for AI infrastructure and the ensuing outcry over a supposed bailout. They distinguish between federal loan guarantees, regulatory reform to enable private buildout, and actual bailouts, ultimately arguing that the latter is neither likely nor requested.
- •Friar’s remark about a federal ‘backstop’ is widely interpreted as asking for a bailout.
- •She quickly walks it back, saying she meant public‑private partnership and industrial capacity, not debt guarantees.
- •Sacks insists there will be no AI bailout: competition is too intense and failure of one lab is tolerable.
- •Policy focus should be on permitting reform, power generation, and ‘behind‑the‑meter’ solutions.
- •Brad praises federal efforts to accelerate nuclear and data‑center buildout without directly taking balance‑sheet risk.
- 25:00 – 36:00
China, Power Constraints, and the Battle Over AI Regulation
Nvidia’s Jensen Huang warns that China is ‘nanoseconds behind’ the U.S. in AI, and the Besties link this to U.S. regulatory fragmentation, power constraints, and Chinese open‑source models gaining traction. Sacks makes a sustained case for federal preemption of state AI rules to avoid ideological capture and competitive disadvantage.
- •Jensen argues China’s centralized buildout and subsidies give it an AI edge; U.S. is slowed by state regs and power.
- •Example: Cursor 2.0 reportedly swaps Anthropic for a cleansed Chinese Qwen‑based model.
- •Sacks: 50‑state AI regulation will default to blue states’ DEI‑infused ‘algorithmic discrimination’ standards.
- •He advocates a single federal framework under Trump to avoid blue‑state ideological dominance in AI.
- •Chamath compares potential AI regulatory patchwork to California’s CARB auto rules distorting the U.S. auto market.
- 36:00 – 46:00
AI Doomerism, Bubble Talk, and Public Backlash Against AI
The conversation shifts to AI’s perception problem: doomer fears about job loss and existential risk, plus anger over higher electric bills, are making politicians afraid to even say ‘AI.’ Sacks links much of the doomer narrative to well‑funded philanthropy, and the group pushes back on both ‘AI bubble’ and ‘superintelligence tomorrow’ memes.
- •Brad warns that public opinion is turning anti‑AI as people blame it for layoffs and power costs.
- •Sacks cites Substack reporting on how a few donors fund dozens of doomer orgs pushing fear narratives.
- •He calls out the contradiction of claiming both ‘AI is a bubble’ and ‘AI is about to become superintelligent.’
- •They argue that if the U.S. slows AI under doomer pressure, China will simply win the race.
- •Consensus: need better ‘antibodies’ against ideological memes that aim to freeze AI progress.
- 46:00 – 55:00
OpenAI vs. Competitors: Revenue Mix, Free Products, and Platform Risk
Jason presses Brad on OpenAI’s sustainability given competition from free products by Google and Apple and startups’ fear of building on a platform that competes with them. Brad responds by emphasizing the size of the AI supercycle, diversified bets across labs, and Anthropic’s more ‘infrastructure only’ strategy that appeals to developers.
- •OpenAI is reportedly on a ~$20B run‑rate, with ~75% from consumer subscriptions.
- •At $20/month, that implies tens of millions of paying users—but Google and Apple can offer free AI via ads.
- •Startups fear OpenAI’s vertical integration and prefer Anthropic, which pledges to avoid the application layer.
- •Brad is long the entire ‘compute stack’ (OpenAI, Anthropic, Google, Microsoft, Nvidia, etc.).
- •He cautions that 3‑year revenue forecasts are guesses; what matters is contract flexibility on capex if growth slows.
- 55:00 – 1:05:00
Macro Check: Consumer Cracks, Delinquencies, and ‘Risk‑Off’ Markets
Brad lays out Altimeter’s macro stance: after a huge rally from early‑year lows, they’ve scaled back to medium exposure as consumer data deteriorates. They discuss rising credit card delinquencies, pressure on low‑income households, strong earnings headlines masking underlying fragility, and the divergence between the mega‑caps and the rest of the market.
- •Consumer is weakening: Chipotle, Cava, cruise lines, and airlines report lower‑end pullback.
- •Credit card delinquencies are back to ~2009 levels; regional banks and credit spreads show stress.
- •Despite this, ~70% of companies are beating earnings, but 2025–26 forecasts look shaky.
- •Chamath notes decoupling: a handful of AI‑rich mega‑caps pull indices up while 400+ S&P names struggle.
- •They argue Trump’s third ‘act’ must focus on tangible domestic wins and middle‑class wealth effects.
- 1:05:00 – 1:16:00
Inflation, Youth Unemployment, and the AI Job-Loss Debate
Jason blames rising youth unemployment and recent layoffs partly on AI, citing startup behavior and leaked Amazon docs, while Sacks pushes back hard, calling it a misdiagnosis driven by anecdotes. The two spar over data on white‑collar employment, wage growth, and whether young grads’ degrees and politics—not AI—are the main problem.
- •Youth (20–24) unemployment is ~9.2% and rising; Jason ties this to entry‑level white‑collar automation by AI.
- •Sacks counters with charts showing white‑collar jobs as a steady share of employment and real wages rising.
- •He notes Amazon’s Andy Jassy explicitly blamed ZIRP/over‑hiring and DEI, not AI, for major cuts.
- •Jason holds his ground based on ‘front‑line’ startup anecdotes but concedes not all layoffs are AI‑driven.
- •Debate highlights the difficulty of isolating AI’s immediate labor impact versus longer‑term structural shifts.
- 1:16:00 – 1:30:00
Filibuster ‘Nuclear Option’ and the GOP’s Domestic Policy Dilemma
With a government shutdown and voters blaming Republicans, Sacks argues the filibuster lets Democrats block Trump’s agenda despite a GOP trifecta. He proposes invoking the ‘nuclear option’ to abolish the filibuster, pass a domestic reform package on AI, crypto, healthcare, and affordability, and face voters on the results in 2026 and 2028.
- •Public blames Republicans for shutdown, even though Democrats can block bills with 41 Senate votes.
- •The filibuster is a Senate custom, not a constitutional requirement, and can be ended with 50 votes.
- •Moderate Democrats (Manchin, Sinema) who protected the filibuster are gone; Dems will scrap it next time they control government.
- •Sacks says Republicans should act first, use their mandate, and deliver concrete reforms.
- •Chamath agrees that without reopening government and legislating, critical domestic agendas (AI, crypto, healthcare, student loans) will stall.
- 1:30:00 – 1:40:00
Mamdani Wins New York: Socialism’s Rise and a Broken Generational Compact
The panel analyzes Zohran Mamdani’s surprise New York City mayoral win as a data point in the rise of left‑wing socialism in blue strongholds. They link his appeal—especially among young women and recent arrivals—to a sense that capitalism is not delivering on housing, student debt, and affordability, echoing Brad’s Stanford ‘ban the billionaires’ experience.
- •Mamdani, a self‑proclaimed democratic socialist, defeats Andrew Cuomo with ~50.4% in an 80/20 Democratic city.
- •Support is strongest among newcomers to NYC; native‑born New Yorkers skew more against him.
- •Brad reports many Stanford students arriving at a debate believing billionaires ‘should not exist.’
- •Friedberg’s earlier prediction: rapid AI‑driven growth plus inequality will fuel socialist movements by mid‑2020s.
- •Brad sees a ‘fight for the soul of America’ over whether the American Dream (economic mobility) still exists.
- 1:40:00 – 1:47:00
Student Debt, Housing, and Why Young Voters Are Turning Socialist
Chamath and Sacks drill into student loans and housing as root causes of generational radicalization. They argue that decades of federally underwritten, mispriced degrees have left young people with ‘negative capital,’ making socialism appealing. Both endorse the idea of tying loan availability to earnings outcomes and contemplate debt relief as part of a major reset.
- •Thiel’s thesis: high student debt plus unaffordable housing produce generations with ‘negative capital’ and no stake in capitalism.
- •Chamath now supports some student loan forgiveness, conditioned on a structural overhaul of how degrees are financed.
- •They propose ending federal underwriting of low‑ROI degrees and letting capital markets price educational risk.
- •Sacks: writing off loans is acceptable only if you simultaneously stop originating bad ones.
- •Housing fix is more local: states like California must reform zoning and permitting that block supply.
- 1:47:00 – 1:58:00
Are Democrats Becoming a Party of Socialist ‘Revolutionaries’?
Sacks warns that the Democratic base’s energy is shifting to figures like Mamdani and Katie Porter, while moderates are purged. He fears a future trifecta where Democrats scrap the filibuster and enact sweeping redistributive and decarceral policies, contrasting this with San Francisco’s recent centrist correction and the policy A/B test it creates versus New York.
- •Sacks describes Democrats’ activist core as ‘genuine socialist revolutionaries’ who despise the capitalist system.
- •New York’s Mamdani wants to close Rikers, eliminate gifted programs, and expand government‑run services.
- •LA County is considering demolishing its only county jail and diverting 7,000 hardened inmates to ‘services.’
- •San Francisco, by contrast, just elected centrist Daniel Lurie to reverse progressive excesses—an A/B test against NYC.
- •If Republicans fail to deliver domestic gains, Sacks fears voters will eventually give socialists full control.
- 1:58:00
Global Startup Pipelines, Angel Investing, and the AI Founder Boom
The show briefly pivots to Jason’s Foundry University expansion to Saudi Arabia and Japan, underscoring the global race to nurture AI startups. They note how early‑stage AI companies are reaching multi‑billion valuations rapidly, and how strong cohort behavior suggests real product‑market fit, even as the macro picture remains choppy.
- •Foundry University launches in Riyadh with 60 startup teams; Japan next, alongside the U.S. program.
- •Partnerships with entities like JETRO and cloud credit offers (e.g., Humane) help bootstrap ecosystems.
- •Jason cites angel bets in AI firms like Cognition, Decagon, and Distill—now at multi‑billion valuations.
- •Cohort curves for several AI SaaS products show users coming back after initial drop‑off—strong engagement signal.
- •They close by reaffirming long‑term conviction in the AI supercycle despite near‑term volatility.
