Skip to content
All-In PodcastAll-In Podcast

Why Anthropic's best model is locked up despite a $30B ramp

Mythos scored so well on cyber offense that Anthropic delayed its release; a 100-day hardening window did not stop Claude Code from driving a $30B run rate.

Jason CalacanishostBrad GerstnerguestDavid SackshostChamath Palihapitiyahost
Apr 9, 20261h 29mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Anthropic’s security hold, agent wars, revenue explosion, and Iran ceasefire debate

  1. The hosts debate whether Anthropic withholding its Mythos model is responsible security practice or fear-based marketing, concluding the cyber-risk is plausible even if the presentation is theatrical.
  2. They unpack the OpenClaw cutoff controversy, framing it as a potential anti-competitive move (or rational repricing) as Anthropic launches first-party managed agents that compete directly with third-party agent frameworks.
  3. They analyze Anthropic’s reported ~$30B revenue run-rate surge as evidence that AI—especially coding/agentic workflows—has crossed a capability threshold with an enormous “TAM for intelligence,” while arguing about how soon profitability matters.
  4. They describe a “major vibe shift” where Anthropic appears to be executing sharply while OpenAI faces narrative and morale pressure, yet several speakers caution that the market is non-zero-sum and OpenAI/Meta/Google remain formidable.
  5. They review the Iran-war ceasefire and regional diplomacy, then argue over whether US involvement reflects Israeli influence versus US sovereign decision-making, emphasizing domestic political risk and Israel’s declining US favorability as a strategic concern.

IDEAS WORTH REMEMBERING

5 ideas

Frontier coding gains translate directly into cyber offense and defense.

As models get better at code, they inevitably get better at finding and chaining vulnerabilities; the group expects a short window where defenders must patch legacy bugs before similar capabilities diffuse broadly (including via China/open source).

A pre-release “sandbox” period can be a pragmatic safety valve without a government moratorium.

Gerstner frames Anthropic’s 100-day coalition approach as industry-led hardening (Apple/Microsoft/Google/etc.) that preserves competitive momentum while reducing immediate exploit risk.

Anthropic’s safety announcements are perceived as partly credible and partly go-to-market muscle.

Sacks argues Anthropic has a pattern of headline-grabbing doom studies, but credits this cyber case as more logically grounded; Chamath calls it mostly theater, citing GPT-2-era precedent.

OpenClaw’s cutoff illustrates how platform owners can reshape markets via pricing and product cloning.

Forcing heavy users off flat-rate subscriptions and onto metered APIs can kneecap third-party agent frameworks; if first-party agents remain effectively “bundled” while competitors pay metered rates, it raises bundling/anti-competitive questions (depending on equivalence and market power).

Coding share today matters because agents likely depend on coding flywheels.

Sacks argues the leader in coding tokens may gain data-scale advantages that carry into agents, making “clean” competitive behavior important before regulators later scrutinize discriminatory pricing or tying.

WORDS WORTH SAVING

5 quotes

Anthropic has proven that it's very good at two things. One is product releases. The second is scaring people.

— David Sacks

I think it's mostly theater.

— Chamath Palihapitiya

We have no choice but to treat it that way.

— David Sacks

It turns out that the TAM for intelligence is radically different than anything that we've seen before.

— Brad Gerstner

If you land the plane on those things... the market could really take off.

— Brad Gerstner

Mythos model withheld and “Project Glasswing” cyber coalitionAI-driven vulnerability discovery and chained exploitsFear-as-marketing vs legitimate frontier riskOpenClaw access cutoff, API pricing, and bundling concernsAgent platforms as next battleground (Claude managed agents, Perplexity, Hermes, etc.)Anthropic ~$30B run rate and enterprise adoption (>$1M ARR customers)Iran ceasefire, Israel-US dynamics, and market reaction

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome