All-In PodcastBig Beautiful Bill, Elon/Trump, Dollar Down Big, Harvard's Money Problems, Figma IPO
CHAPTERS
- 0:00 – 7:10
Summit Hype, Tequila Launch, and Personal Trips
The hosts open with banter about the upcoming All‑In Summit, JCal’s tequila brand, and recent trips to Vegas and the Snake River. The tone is light, setting the stage before they pivot into policy and markets.
- •Promotion of the All‑In Summit and tequila.allin.com
- •JCal and Friedberg’s Vegas weekend versus JCal’s family rafting trip
- •Casual discussion of Snake River, wagyu beef, and travel preferences
- 7:10 – 13:20
The ‘Big Beautiful Bill’ and AI Federal Preemption Fight
They break down the Big Beautiful Bill’s passage in the Senate, focusing on the removal of federal preemption for AI regulation. Friedberg and Chamath argue that AI should be regulated nationally due to its cross‑border nature and national security implications.
- •BBB passes Senate with J.D. Vance’s tie‑breaking vote; must return to House
- •10‑year moratorium on state AI regulation removed; Cruz failed to get a 5‑year compromise
- •Over 70 AI‑related state laws passed and 1,000+ bills proposed in 2025
- •Friedberg: patchwork AI laws will cripple internet‑scale providers and harm consumers and jobs
- •Chamath: AI is akin to defense; state‑level AI law is as nonsensical as state armies
- •Risk that complex compliance regimes entrench incumbents and shut out startups
- 13:20 – 19:40
States’ Rights, Self‑Driving, and Where AI Regulation Belongs
The hosts compare AI regulation to issues like abortion, drugs, gambling, and self‑driving cars to test where federal vs state authority should sit. They acknowledge that some AI‑related issues, like local road safety, may warrant state input, but argue that model regulation itself is beyond local governments’ technical competence.
- •Chamath distinguishes AI from abortion: one is individual choice, the other national security
- •Jason uses gambling and cannabis as analogies for state‑by‑state experimentation
- •Debate over whether states should control self‑driving deployments on local roads
- •Consensus that cities/states lack expertise to regulate AI models themselves
- •Foreshadowing of how energy and AI are linked in the bill
- 19:40 – 28:00
Energy Subsidies, Nuclear Hopes, and Powering the AI Boom
Focus shifts to the bill’s energy provisions, including removal of certain green subsidies and EV tax credits. Friedberg lays out a case for massive electricity expansion, possibly led by nuclear, while Chamath discusses real‑world constraints from his new data center project.
- •IRA green incentives for solar, wind, and EVs are cut back in the BBB
- •Secretary of Energy argues existing gas, oil, coal, and nuclear capacity is enough; less need to subsidize renewables
- •Friedberg: dependence on subsidized energy is unsustainable; we need scalable, market‑driven sources
- •Hope that subsidy removal spurs nuclear investment, aided by new deregulation EOs
- •Chamath’s 1 GW data center near a nuclear plant outside Phoenix illustrates demand for firm power
- •Key constraints now are supply chain, transmission, and local/state permitting, not generation technology
- •Both stress: do nothing that slows electricity production or safe storage (batteries); AI and GDP depend on it
- 28:00 – 38:40
All Energy Is Good Energy: Coal, Solar Economics, and Grid Realities
Chamath admits he nearly bought a coal company and would invest across all energy types, while Friedberg explains why solar remains attractive despite subsidy risk. They note that grid utilization and downstream constraints often matter more than simple generation capacity metrics.
- •Chamath regrets not buying a coal company that later quintupled; advocates being energy‑agnostic
- •Solar’s key advantage: ~17‑month payback from investment to revenue, versus decade‑long nuclear timelines
- •California grid averages ~45–50% utilization, with only about five days of true energy deficit per year
- •The choke points are distribution, storage, and peaks, not necessarily base generation in normal conditions
- •Friedberg advocates stripping all subsidies but removing barriers to any new electricity production
- 38:40 – 49:20
Elon vs. Trump: Spending, Deficits, and the Tech–MAGA Alliance
The hosts unpack the public spat between Elon Musk and Donald Trump over the bill’s spending and deficits. Friedberg argues both sides have valid points on fiscal policy but warns that tech and MAGA are now mutually dependent and cannot afford a lasting conflict.
- •Elon denounces the bill as “insane spending” and calls America a one‑party “Porky Pig” state
- •Trump previously joked about deporting Elon and “sicking Doge” on him over subsidies
- •Friedberg: Elon is right to highlight the debt spiral and government inefficiency; Doge project attempted to address this
- •White House argument: this bill mainly tweaks mandatory spending, with discretionary cuts and impoundment to come later
- •Tariffs (e.g., 20% on Vietnam) are projected to boost revenue; CBO doesn’t fully model dynamic effects of tax cuts on GDP
- •Core unresolved debate: do tax cuts stimulate broad growth or just enrich the wealthy?
- •Friedberg: MAGA needs tech’s innovation; tech needs MAGA’s regulatory stance on AI—open conflict is risky for both
- 49:20 – 59:20
Debt Spiral, Absolutist Politics, and the 3‑3‑3 Target
Zooming out, Friedberg cites Ray Dalio and Balaji to describe a late‑empire dynamic where political incentives make cutting spending nearly impossible. He lays out a 3‑3‑3 framework—deficit, growth, inflation—as a theoretical path to stability, but acknowledges skepticism that Washington can deliver.
- •Dalio sees low odds that U.S. alters its debt trajectory; absolutist politics block compromise
- •Congressional incentives: bring more federal money home to districts; White House: deliver campaign promises
- •Current snapshot: ~6% deficit/GDP, ~1.4% GDP growth, ~2.4% inflation
- •Goal: 3% deficit, 3% growth, 3% inflation; AI and productivity are key to hitting the growth number
- •Balaji: real debt includes federal, state/local, corporate, consumer, and unfunded pension liabilities
- •Friedberg: once society becomes dependent on high spending, it’s politically impossible to reverse; inflation/printing likely outcome
- •Warning that over‑regulating AI could choke the very growth engine needed to escape the debt trap
- 59:20 – 1:14:40
Dollar Down Big: Trade, Inflation, and Asset Markets
They examine the dollar’s sharp drop in 2025 and debate whether it’s a crisis or a continuation of a decades‑long trend. Friedberg emphasizes the immediate inflationary squeeze on imports and consumers, while Chamath focuses on the relative attractiveness of U.S. assets and human capital.
- •Dollar down ~11% YTD; worst start in 50+ years; down ~6% from pre‑election levels
- •Friedberg: 4–5T in imports just got ~11% more expensive in dollar terms, independent of tariffs
- •Rising living costs without wage growth fuel political demand for “free” government services and socialism
- •Chamath: dollar has gradually devalued ~50% over 35–40 years; trend is longstanding and manageable
- •As long as U.S. assets (equities, real estate, hard assets) outperform currency decay, investors remain ahead
- •Foreign holdings of Treasuries have fallen from ~34% to ~22% in 10 years—Chamath views this as positive (less foreign leverage over U.S. policy), Friedberg as a warning on who will finance U.S. debt
- •Both see enormous global cash piles and ongoing “flight to quality” into U.S. assets, anchored by American innovation
- 1:14:40 – 1:41:40
Harvard’s Money Problems and the Coming University Reckoning
The conversation pivots to Harvard’s deepening conflict with the Trump administration over antisemitism, DEI, and federal funding, using it as a case study of how elite universities face both political and structural threats. Chamath predicts forced asset sales and financial pain, while Friedberg argues that AI and the internet are dismantling the traditional higher‑ed monopoly.
- •Trump admin canceled >$2B of Harvard research grants and is threatening civil rights action over antisemitism
- •Potential conditions: cancel DEI initiatives, third‑party oversight of admissions, mandatory antisemitism measures
- •WSJ reports Harvard would face ~$1B annual budget shortfall if cuts and taxes proceed
- •Chamath: Harvard’s endowment has ~40% in private equity, doubled its PE allocation into peak valuations; will be forced to sell at steep discounts (20–40%) if cash‑strapped
- •New BBB excise tax on foundations could be as high as 4–8% annually on endowment assets
- •Friedberg: universities have two functions—education and research—both are being structurally disrupted
- •Internet democratized access to lectures (e.g., MIT OpenCourseWare); AI can deliver personalized, near‑free tutoring to anyone globally
- •Independent research institutes globally show research doesn’t need to be bundled with campus education
- •Long‑term: prestige brands may lose monopoly as employers adopt new filters (coding challenges, project portfolios, in‑house training) and as degree logos decouple from performance
- 1:41:40 – 1:58:00
Figma IPO, Superhuman Acquisition, and AI’s Pressure on SaaS
The hosts end by surveying a hot IPO/M&A market—Figma’s S‑1, Grammarly–Superhuman, and others—against the backdrop of explosive AI model revenue forecasts. Chamath warns that foundational models may absorb large swaths of software functionality, creating uncertainty about the long‑term durability of many SaaS revenues.
- •Grammarly acquires Superhuman; Grammarly also bought Coda, building an AI workplace suite
- •Figma S‑1 highlights: $228M Q1 revenue, 13M MAUs, $1.5B cash, no debt, ~40% growth, strong cash flow and net retention
- •CoreWeave and others have gone public; IPO/M&A market is warming amidst expectations of Fed rate cuts
- •Chamath cites internal forecasts: OpenAI at $13B revenue in 2025 and $125B by 2029; Anthropic at ~$35B by 2027
- •Meta offering $300–500M comp packages signals perceived existential threat from AI to traditional social and software models
- •Key investor question: are tools like Figma adjacent to, or subsumed by, core AI model workflows?
- •Chamath suggests a relative‑value trade: long Figma, short Adobe to capture spread while hedging AI model risk
- •Jason and Friedberg discuss “revenue durability” as a critical lens: will buyers and even entire departments still exist if AI agents automate their function?
- 1:58:00
Wrap-Up: Culture, Poker, AI, and Media Recommendations
They close with lighter banter about poker, possible gambling changes in the bill, and media recommendations, including Bob Dylan albums and sci‑fi films. It underscores how their discussions blend hard policy with culture and personal interests.
- •Poker talk, All‑In “ledger,” and potential regulatory impacts on gambling professions
- •Chamath studying Modern Poker Theory to improve his game
- •Discussion of logistics/tax complexity around private poker ledgers if laws change
- •Recommendations: Denis Villeneuve’s ‘Arrival,’ classic sci‑fi like ‘Logan’s Run’ and ‘Silent Running’, and several Bob Dylan albums (‘Blood on the Tracks,’ ‘Infidels,’ ‘Empire Burlesque,’ ‘Street Legal’)
- •Reflection on Dylan, Joan Baez, and musical influence, tying back to broader cultural commentary
