TLDR
The All-In Podcast hosts discuss Anthropic's decision to withhold its Mythos model due to cybersecurity risks, the company's explosive revenue growth to $30B run rate, competition among frontier AI labs, the Iran ceasefire negotiations, and X's new auto-translate feature. Brad Gerstner joins as guest to provide investor perspective on Anthropic's trajectory and the broader AI landscape.
For you
PRO members can sign in and request more information — or their own version of this digest, tuned to what they care about.
Sign in to unlockSurprising Insights
Anthropic achieved $30B revenue run rate with only ~2,500 employees and 1.5-2 gigawatts of compute — far fewer resources than Google needed at similar revenue scale.
X's auto-translate feature is surfacing real-time translated posts from Japanese, Hebrew, Russian, and Chinese users directly into English feeds, enabling cross-language discourse at unprecedented scale.
Anthropic cut off OpenClaw (the #1 open-source project on GitHub) from using $200/month subscriptions, forcing users to expensive API pricing, just before launching their own competing agent product.
Key takeaways
Anthropic withheld its Mythos model after discovering it could autonomously find and chain vulnerabilities across major systems, launching Project Glasswing with 40 major companies for a 100-day hardening period.
Anthropic's revenue run rate exploded from $1B to $30B in roughly 18 months, driven by enterprise adoption of Claude Code, making it the fastest revenue ramp in tech history.
The panel debates whether Anthropic's safety messaging is genuine or marketing theater, with Chamath calling it 'mostly theater' while Sachs and Brad see legitimate cybersecurity concerns.
OpenAI is preparing to release its 'Spud' model (Blackwell-trained) imminently, and the panel agrees the AI race is non-zero-sum with a massive TAM for intelligence.
The Iran ceasefire negotiated by Trump's team (Vance, Kushner) appears to be holding, with markets pricing in de-escalation and potential for broader Middle East normalization.
Notable quotes
“The model that we're experimenting with is by and large as good as a professional human identifying bugs. It has the ability to chain together vulnerabilities... this model is able to create exploits out of three, four, sometimes five vulnerabilities that in sequence give you some kind of very sophisticated end outcome.”
▸ Notable quotes · 5:19“We're seeing the fastest revenue growth in history. It's utterly unprecedented. And this is just one category or vertical of AI. We know that agents are coming next and the enterprise adoption of that is going to be absolutely massive.”
▸ Notable quotes · 56:40“The unit cost of intelligence is plummeting. Not the cost of tokens. The unit cost of intelligence is plummeting because the capabilities of these models is so much better.”
▸ Notable quotes · 47:26“This is not zero sum. The TAM of intelligence is dramatically larger than any TAM we've ever seen in our investing careers over the last two decades.”
▸ Notable quotes · 60:38Chapters
4:00 — Anthropic Withholds Mythos Model Over Cyber Risks: Anthropic announces it's withholding its Mythos model after discovering it can autonomously find and chain vulnerabilities across major operating systems and browsers. They launch Project Glasswing with 40 major companies for a 100-day hardening period.
9:00 — Debate: Safety Theater vs. Legitimate Threat: Panel debates whether Anthropic's move is genuine safety precaution or marketing theater. Chamath calls it theater citing GPT-2 precedent; Sachs and Brad see legitimate cyber risk requiring industry coordination.
24:00 — OpenClaw Cutoff and Competitive Dynamics: Anthropic cut off OpenClaw users from $200/month subscriptions, forcing API pricing, shortly before launching their own competing agent product. Panel discusses bundling, antitrust, and open source vs. proprietary dynamics.
43:00 — Anthropic's Explosive Revenue Growth: Anthropic's revenue run rate hit $30B (from $1B 18 months ago), driven by Claude Code enterprise adoption. Brad projects $80-100B exit run rate by year-end. Discussion of gross margins, compute constraints, and profitability timeline.
55:00 — AI Competition Landscape and Value Capture: Analysis of competition among Anthropic, OpenAI, Meta, Google, xAI. Discussion of value capture across stack layers (chips, hyperscalers, models, applications) and implications for enterprise software.
70:00 — Iran Ceasefire and Middle East Geopolitics: Trump team (Vance, Kushner) negotiating ceasefire in Islamabad. Markets show minimal drawdown, pricing in de-escalation. Discussion of Israel-US relations, Netanyahu's influence, and potential for broader regional normalization.
82:00 — X's Auto-Translate Feature and Global Discourse: X's new auto-translate surfaces real-time translated posts from Japanese, Hebrew, Russian, Chinese feeds. Panel praises it as breakthrough for cross-cultural understanding and potential Nobel Peace Prize-worthy.
Speakers
Jason Calacanis — Host
Chamath Palihapitiya — Co-host/Investor
David Sacks — Co-host/Investor
Brad Gerstner — Guest/Investor (Altimeter Capital)
Brian Gerstner — Mentioned as fifth bestie