All-In PodcastThe Stablecoin Future, Milei's Memecoin, DOGE for the DoD, Grok 3, Why Stripe Stays Private
CHAPTERS
- 0:00 – 13:00
Intro, Lost Stripe Investment Story, and Stripe’s Scale Today
Jason welcomes the Collison brothers and recounts Chamath’s missed chance to invest in their pre-Stripe startup. The group sets context on Stripe’s growth from a startup payments API to global infrastructure processing over $1 trillion annually, now serving not only startups but large enterprises.
- •Chamath recalls meeting the teenage Collisons at their early startup Automatic and failing to invest, calling it a billion-dollar mistake.
- •Stripe now processes more than $1T per year, roughly 1% of global GDP, for both startups and large enterprises like Amazon, Ford, and Hertz.
- •Stripe has expanded from core payments into lending, card issuing, treasury, cross-border money movement, and more.
- •Stripe’s value proposition evolved as they realized payments were “broken for everyone,” not just startups.
- 13:00 – 21:00
Stablecoins, Bridge Acquisition, and Challenging Card Networks
John and Patrick explain Stripe’s long engagement with crypto, why Bitcoin failed as a payment method, and why modern stablecoins on efficient chains now matter. They outline how Stripe integrates USDC via Bridge and why stablecoins’ primary utility is cross-border and for non-US users, while unpacking the economics of Visa/Mastercard and interchange.
- •Bitcoin was tried on Stripe as a payment method but was too slow, costly, and volatile for transactions; it works better as a “gold substitute” than as payments rail.
- •Stablecoins on Ethereum L2s/Solana now have low fees and fast finality, enabling practical payment and treasury use cases.
- •Stripe acquired Bridge, “the Stripe of stablecoins,” to provide APIs for stablecoin flows; All-In could turn on USDC acceptance with a switch in Stripe.
- •The most compelling early stablecoin use cases are cross-border B2B payments, remittances, and dollar access in high-inflation markets.
- •Patrick explains that most interchange funds bank lending and consumer rewards, complicating simple “kill Visa/Mastercard” narratives.
- 21:00 – 35:00
Dollar Supremacy, Eurodollars 2.0, and Stripe’s Network Effects
The conversation explores how regulated US-backed stablecoins reinforce dollar dominance and parallel the eurodollar system, but with access down to $10 balances globally. They also discuss network effects within Stripe’s ecosystem and the potential for lower fees when both sides of a transaction are on Stripe.
- •Patrick likens stablecoins to a consumer-scale version of the eurodollar system, giving individuals in countries like Ecuador and Nigeria access to dollar savings.
- •US-regulated stablecoins (USDC, future Stripe-linked coins) force issuers to hold Treasuries, which in turn supports U.S. dollar supremacy.
- •All-In could accept USDC through Stripe today; outbound stablecoin payments are technically possible, but domestic bank transfers remain “good enough” so focus is on cross-border use.
- •Stripe sees growing volumes where both counterparty accounts are on Stripe, opening opportunities to treat movements as ledger entries, cut fraud, and eventually reduce costs.
- •Fraud and KYC/ID verification are large cost centers; Stripe positions itself as a reputation and risk network, not just a dumb pipe.
- 35:00 – 46:00
Economic Data from Stripe and Rethinking Financial Infrastructure
Chamath asks whether Stripe could publish economic sentiment data, given its transaction visibility. Patrick describes attempts to derive leading indicators for inflation and explains the difficulty of making Stripe’s hyper-growth data representative. Friedberg then asks how a financial system would be built from scratch today, prompting discussion on instant payments, trust, and fraud.
- •Stripe’s data could support better macro indicators (e.g., inflation leading indicators), but skew toward online and high-growth sectors makes broad extrapolation tricky.
- •Stripe is intellectually open to publishing more aggregated macro data as a public good when methodology is robust.
- •In a “from scratch” financial system, John stresses real-time, scalable payment capabilities and embedded trust/fraud controls as design primitives.
- •Stripe’s network sees 93% of cards previously, enabling fraud detection via familiarity with cards, emails, phones, and device patterns.
- •Fraud is getting worse industry-wide due to ML, AI, deepfakes, and globalization, making the reputation layer increasingly critical.
- 46:00 – 1:00:00
Jamie Dimon’s Anti-Zoom Rant, Remote Work, and Meeting Bloat
Playing Jamie Dimon’s tirade against Zoom multitasking and remote work, the hosts and Collisons discuss Stripe’s hybrid approach and productivity data. They debate leadership tone, the post-woke shift to “leading from the front,” the cost of bloated committees, and tactics like Shopify’s mass meeting purge.
- •Stripe had offices pre-COVID and still mostly operates office-centric with remote pockets; their pre-pandemic data shows remote is bad for early-career productivity and mental health.
- •Friedberg frames Jamie Dimon as part of a broader move away from “coddled workforces” toward more direct, front-led leadership (Zuckerberg, Elon, Coinbase).
- •Jamie Dimon’s view that every department can be 10% leaner leads into a discussion of bureaucratic bloat, over-committee’d decisions, and software-fragmented orgs.
- •Chamath argues that off-the-shelf SaaS (Salesforce, Workday, etc.) hardens rigid role boundaries and promotes bureaucracy instead of true productivity.
- •Tobi Lütke’s script to delete all recurring meetings at Shopify is cited as evidence that many meetings are inertia, not necessity.
- 1:00:00 – 1:12:00
Performance Management, ‘Mids,’ and Heterogeneous Work Models
Jason presses Patrick on how Stripe handles “mids” and low-performers. Patrick emphasizes aggressive but humane performance management and warns against moralizing about office vs remote, noting that different companies and roles legitimately require different setups.
- •Stripe aims to avoid equilibrium where low-performers linger; it’s bad for both the company and the individual’s sense of success.
- •Patrick stresses empiricism over ideology: some world-class companies (Nvidia, Coinbase, Shopify) are remote-friendly; others (Jane Street) insist on co-location.
- •Remote work suitability varies by person, role, and industry; singular “shoulds” around office presence are misleading.
- •Labor productivity in the U.S. is up ~20% over 10 years, contradicting narratives that remote/hybrid inherently killed productivity.
- 1:12:00 – 1:33:00
Defense Spending, Tech-First Doctrine, and Procurement Reform
The group tackles proposed 8% annual cuts to U.S. defense spending and the Doge series’ promise to scrutinize big budget items. Chamath and Friedberg argue that emerging technologies like drones and AI should reshape the force structure and that the U.S. may be shifting into a multipolar worldview where primacy gives way to stable shared power.
- •Chamath argues defense budgets must be set after understanding tech capabilities; spending trillions on frigates while startups build cheap autonomous systems is misaligned.
- •Friedberg describes a strategic transition from U.S. primacy to multipolarity (U.S., China, Russia), with implications for reducing global defense spending and shifting to deterrence.
- •Drones in Ukraine demonstrate asymmetric value: cheap systems destroying premium assets; China’s drone factories can produce millions per month.
- •Patrick notes the proposed cuts resemble historical drawdowns and that critiques of defense procurement inefficiency are long-standing and bipartisan.
- •Book recommendation: “Boyd” by Robert Coram, about Air Force reformer John Boyd and the fight to design better aircraft against institutional resistance.
- 1:33:00 – 1:54:00
Milei’s Libra Meme Coin Scandal and the Ethics of Meme Coins
The show shifts to Argentine President Javier Milei’s disastrous endorsement of the Libra meme coin, which pumped then crashed, triggering lawsuits and an impeachment push. The group uses this as a case study in leadership failure and the broader issue of meme coins as turbocharged gambling products.
- •Milei promoted Libra as a project to grow Argentina’s economy; it hit a ~$4B market cap, then collapsed ~95%, wiping out many of his own supporters.
- •Chamath criticizes Milei’s weak denial, possible family involvement, and the predictable rug-pull dynamics, saying the coverup is worse than the original mistake.
- •Friedberg calls meme coins largely unproductive gambling, akin to collectibles but with friction removed and social virality magnified.
- •Patrick and John liken meme coins to gambling/sports betting: in principle allowed, but with severe negative EV and many heartbreaking whale-loss stories.
- •John highlights state lotteries and legalized sports betting as part of the same social challenge: governments and platforms aggressively marketing negative-EV products to vulnerable people.
- 1:54:00 – 2:11:00
Friendship, Politics, and the Missing ‘Great Debate’ in Public Life
John reflects on the appeal of All-In as a show about genuine friendship, comparing it to ensemble TV like Friends and The West Wing. The group then touches on West Wing vs House of Cards depictions of politics and the lost ideal of a ‘great debate’ focused on national interest rather than tribal point-scoring.
- •John suggests All-In’s appeal mirrors shows where tight-knit friends argue, trust, and joke, making disagreements watchable.
- •Friedberg praises The West Wing’s depiction of substantive policy debates and laments modern politics’ fixation on attacking the other side rather than reasoning about national interest.
- •Patrick counters that The West Wing still reflected a single ideological worldview, and that the 1990s Clinton era, though prosperous, wasn’t a time of broad ideological pluralism.
- •They note how real governance, per Dominic Cummings and The West Wing episodes, involves constant distraction from long-term agendas by near-daily controversies.
- 2:11:00 – 2:30:00
Arc Institute, EVO2 Biology Model, and AI for Complex Disease
Patrick introduces the Arc Institute, a nonprofit basic-research organization he co-founded, and its new open-source EVO2 foundation model trained on genomic data. They discuss how curiosity-driven research can escape NIH grant constraints, and how new tools might finally crack complex diseases like Alzheimer’s, cancers, and autoimmune disorders.
- •Arc funds scientists with flexible, curiosity-driven support, freeing them from restrictive, consensus-based NIH grant patterns that discourage risk and cross-field work.
- •Arc’s hypothesis: single-cell sequencing, functional genomics/CRISPR, and modern transformers create a new ‘read–think–write’ loop capable of tackling complex diseases.
- •EVO2 is trained on 9 trillion base-pair tokens of DNA, mostly from non-human species; despite seeing only one human genome, it achieves state-of-the-art performance in predicting pathogenic human mutations like BRCA variants.
- •The model is fully open source, including training code, and can be fine-tuned for specific tasks such as structure prediction and variant interpretation.
- •Arc is donor-funded (~$100M/year, ~230 people), with potential future self-sustainability via spinouts, but it’s structurally a nonprofit with no financial upside for the Collisons.
- 2:30:00 – 2:42:00
Biotech Futures: From Plant Traits to Engineered Organisms
The panel explores practical biotech applications, from saving diseased tree species to engineering resilient crops. Friedberg describes his company Ohalo’s work on gene-editing traits like disease resistance in strawberries and bananas, and the long-term vision of software-specified organisms tailored to harsh environments like Mars.
- •John raises concern over global tree die-offs (Irish ash, American chestnut, bark beetles) and asks how biotech can help restore species.
- •Friedberg explains trait programs: identify genomic changes (e.g., silencing immune suppressors) that improve disease resistance, then regenerate and propagate trees or crops.
- •Ohalo is partnering with University of Florida to introduce resistance against a major fungal pathogen in strawberries; similar approaches apply to bark beetles and other threats.
- •Bananas highlight the cost of low genetic diversity: nearly all commercial plants are clones of a single Cavendish line, making them vulnerable to TR4 fungus and requiring heavy fungicide use.
- •The aspirational end state: software that, given a desired phenotype (e.g., a plant that thrives on Mars), can generate the required genome—though phenotypic data and multi-scale modeling are still major bottlenecks.
- 2:42:00 – 3:01:00
Grok-3, Colossus, and the Power of Constraints in AI Infrastructure
Chamath uses Elon Musk’s Grok-3 and the Colossus data center as an example of how constraints drive innovation and why he’s become more bullish on pre-training at scale and Nvidia. He recounts the extreme engineering required to stand up 100,000+ GPUs in 122 days and draws parallels to historical industrialists like Henry Kaiser.
- •Chamath admits he was previously skeptical that bigger clusters would keep delivering returns; Grok-3’s performance suggests substantial remaining gains from larger-scale pre-training.
- •Colossus required finding a massive power-capable site (an old Electrolux factory in Memphis), scaling from ~15 MW to ~250 MW, buying a third of U.S. portable liquid cooling capacity, and rewriting Tesla Powerpack firmware for power smoothing.
- •He argues true innovation always has a hard constraint: if capital and talent are abundant, time must be the rigid constraint to force breakthroughs.
- •The “keiretsu” nature of Elon’s companies—Tesla engineers working on xAI infrastructure—resembles Japanese conglomerate structures, enabling fast cross-company problem-solving in atoms, not just bits.
- •Jason shows benchmarks where Grok-3 outperforms leading LLMs on math, science, and coding tests, though the panel notes benchmark interpretive caveats.
- 3:01:00 – 3:13:00
Asteroid Risk, Planetary Defense, and AI-Enabling ‘Impossible’ Projects
Friedberg walks through a newly tracked asteroid’s size, impact energy, and updated probability of hitting Earth in 2032. The group compares it to the 1908 Tunguska event and reflects on how AI could enable complex planetary-defense projects that would once have been intractable.
- •The asteroid is estimated at 80–320 feet in diameter; at the high end, an impact would approximate a 20-megaton blast—devastating regionally, but not an extinction event.
- •Updated estimates place the impact probability at ~1.5%, and dropping quickly as new observations refine its trajectory; likely overall city-hit odds are in the basis-point range.
- •Tunguska (1908) produced an airburst equivalent to ~1,000 Hiroshima bombs, flattening ~80 million trees in Siberia but causing minimal recorded human casualties.
- •Interception would require precise trajectory modeling, launch, and rendezvous timing—technically very challenging but the reason NASA funds planetary-defense tracking.
- •Friedberg posits AI will unlock previously infeasible mega-projects (deep-earth mining, off-world colonization, planetary defense) by coordinating large, complex operations at scale.
- 3:13:00 – 3:31:00
Stripe’s Profitability, Long-Term Vision, and Rationale for Staying Private
Jason finally asks the Collisons when Stripe will IPO. They respond that Stripe is already GAAP-profitable and treat being public as a financing tactic, not a virtue signal, emphasizing durability, long-term compounding, and freedom from short-term market noise.
- •Stripe is GAAP-profitable on a fully loaded net income basis (no adjusted-EBITDA games) while processing >$1T in annual volume.
- •John rejects the idea that public markets are needed for discipline: if management requires a 25-year-old analyst to stay focused, “something is horribly wrong.”
- •Being public is defined as access to cheaper, deeper, more liquid capital; it’s not morally superior and comes with volatility that can hurt employees who join at peaks.
- •Financial services historically stay private for long periods (Bloomberg, Fidelity, Vanguard, Jane Street; Goldman ~130 years, Visa ~50 years before IPO).
- •The brothers say Stripe is their life’s work; the decision framework is: what maximizes 10–20-year compounding and customer value, not what optimizes for an IPO pop.
- 3:31:00
Closing Banter, Media Plans, and Jeopardy Tease
The episode winds down with a recap of themes—work discipline, constraints, and leadership—plus promotional notes about All-In’s upcoming South by Southwest event and Friedberg’s appearance on Celebrity Jeopardy. The Collisons join the lighthearted ribbing, reinforcing the friendly dynamic that underpins the show.
- •Jason summarizes tongue-in-cheek lessons: wear pants and go to work; constraints create art; South American presidents should avoid meme coins.
- •All-In announces a live event at SXSW with Jason and Friedberg on media and business-building, powered by Stripe ticketing.
- •Friedberg teases his Celebrity Jeopardy appearance, hinting at buzzer and format challenges rather than pure knowledge deficits.
- •The episode closes on camaraderie, with repeated jokes about mispronouncing “Collison,” defusing earlier sharp critiques, and reinforcing the show’s friendship-first vibe.
