All-In

Chamath Palihapitiya × Jason Calacanis × David Sacks × David Friedberg

Open-source token usage flipped from 20% to 80% in 12 weeks; if more than 60–70% of frontier tokens are replaceable, Anthropic’s S-1 revenue must be discounted.Meta’s free Muse hit #1 in the App Store with 3M downloads in roughly 10 days, challenging the App Store’s 30% rev-share as agents make services headless.Converging models push OpenAI and Anthropic toward higher-value applications, while their six-to-12-month frontier lead remains the central risk.

Dina Powell McCormick

Meta’s multibillion-dollar Richland Parish project lifted sales-tax growth from 5–10% to a 260% peak and certified employees’ checks from $10,000 to $50,000.Louisiana shifts generation, grid resilience, upgrades and storm costs to Meta; the Academy produced 250 graduates and 90% retention from 40,000 applicants, while Google and BlackRock support scaling amid backlash and Cotton and Warner’s adversary theory.

Chamath Palihapitiya × Jason Calacanis × David Sacks × David Friedberg

Anthropic’s IPO faces a disclosure test after Evan Hubinger endorsed a greater-than-10% extinction probability, despite Polymarket’s 88% odds.Sacks argues rapid amplification may reflect a coordinated campaign targeting open-source weights, though that remains his interpretation.OpenAI’s Navier–Stokes result shows brute-force leverage, while data-leakage risks favor sovereign deployment and Nike’s 80% decline underscores a product, channel, and brand reset.

ChatGPT-6 (“Astra ASI”) is rolling out to limited organizations as the panel treats AGI as present, with rivals potentially catching up within “the next 3 or 4 months.”Falling intelligence costs may make non-frontier models commodities, while real revenue fuels 1998-style euphoria; Nvidia’s prospective Hugging Face transaction could bolster open competition through cheaper sovereign compute, with regulation a key risk.

Chamath Palihapitiya × Jason Calacanis × David Sacks × David Friedberg

Anthropic’s regulatory fight is widening into bipartisan data-center pushback as rising yields compress risk capital.Chamath says frontier labs are most exposed because non-investment-grade balance sheets cannot fund the compute required for growth.Sacks predicts an “AI FINRA” could exclude open models through uniform standards, while Friedberg says recursive self-improvement is likely to happen soon.

Gavin Baker

Anthropic’s reported $2T IPO figure may be banker theater rather than a clearing price, while Gavin Baker and Sacks estimate 2027 exit ARR at roughly $400–500B, constrained by physical supply.NVIDIA’s $500B financing effort could make GPU compute an asset class and support a capital-light cloud business, while Anthropic’s S-1 will test demand and dark-GPU risk.

Chamath Palihapitiya × Jason Calacanis × David Sacks × David Friedberg

A US ban on Chinese open-source models could impose a 50-100x AI input cost premium on domestic developers while failing to stop distillation, creating a policy-driven “token tax.”With models reportedly matching published performance within weeks, 95% of real work may be served by many models, shifting durable value toward applications, clouds, and chips while Anthropic’s $1.5 billion settlement leaves training legality unresolved.

Chamath Palihapitiya × Jason Calacanis × David Sacks × David Friedberg

Demis Hassabis’s FINRA-style SRO emerged as the least-bad AI oversight option, but startup and open-source representation, catastrophic-risk limits, voluntary proof and preemption remain essential.A reported $53 billion Stripe-led PayPal bid could combine 439 million consumer accounts with Stripe’s merchants, Braintree, Venmo, Cash App and stablecoin infrastructure to challenge Visa and Mastercard.Electricity is the binding constraint: a 7-8 GW PGM auction drew roughly 156 MW, while New York’s moratorium makes energizable sites scarcer and deployment a catalyst to monitor.

Andrew Feldman × Robin Rombach

Cerebras reports a $25 billion backlog as faster inference turns token capacity and data-center execution into scarce product advantages, while open models take routine and sovereign workloads.Feldman expects its architecture to improve well over 2x in 18 months; staged releases, red-teaming and recursive model improvement remain catalysts—and risks—as multimodal AI reaches film and robotics.

Chamath Palihapitiya × Jason Calacanis × David Sacks × David Friedberg

Palantir and NVIDIA’s sovereign-AI pitch shifts enterprise value toward owning compute, data, model weights, and proprietary “alpha” rather than renting frontier intelligence. 8090 made Claude 1.4x to 4x cheaper and 1.5x faster than Opus 4.8 alone, while a 16.4x-cheaper open model and falling token prices could challenge an OpenAI-Anthropic duopoly, with AI’s employment impact still unresolved.