
Anthropic's $30B Ramp, Mythos Doomsday, OpenClaw Ankled, Iran War Ceasefire, Israel's Influence
All-In with Chamath, Jason, Sacks & Friedberg
Hosted by Chamath Palihapitiya, David Sacks and Jason Calacanis
- 1h 29m
- guest
- 1
- Transcript
Summary
Written by a model from the transcript; not yet reviewed.
Brad Gerstner fills in for David Friedberg as the group debates Anthropic's decision to withhold its Mythos model after it autonomously found thousands of vulnerabilities, including a 27-year-old bug in OpenBSD and a 16-year-old FFmpeg flaw missed after 5 million automated scans. Sacks gives Anthropic credit this time, noting the logical connection between coding capability and vulnerability detection, while Chamath calls it 'mostly theater' by comparing it to GPT-2's 2019 rollout. Gerstner, an investor in both Anthropic and OpenAI, argues the self-regulation through Project Glasswing—40 companies spending 100 days hardening systems—shows market forces can handle safety without government mandates.
The conversation shifts to Anthropic's revenue ramp from $1B at end of 2024 to $30B by April 2026, with over 1,000 enterprises paying over $1M annually. Gerstner predicts an $80-100B exit run rate by year's end, noting the company is compute-constrained at 1.5-2 gigawatts. They also dissect Anthropic's move against OpenClaw, where the company cut off $200 subscription access for power users and released a competing agent product within 10 days, raising antitrust questions about bundling and price discrimination.
The episode closes with the Iran ceasefire, where Trump agreed to a two-week truce if Iran opens the strait, with Vance and Kushner heading to Islamabad for talks. The group discusses Israel's influence on US foreign policy, with Chamath arguing the risk is more problematic for Israel than the US, and Jason praising X's auto-translation feature as the most impressive tech release in years.
- Anthropic's Mythos model found a 27-year-old vulnerability in OpenBSD and a 16-year-old FFmpeg bug that survived 5 million automated scans, demonstrating that AI coding models can uncover dormant security flaws at scale. 4:47
- As coding models become more capable, they will be more capable of finding and chaining vulnerabilities, creating a one-time window where AI-driven cyber defense can detect dormant bugs before they are exploited. 11:52
- Chamath argues that if these exploits are this easy to find, patching them all would require shutting down the internet for about five years, making the 100-day timeline unrealistic for full remediation. 16:09
- Anthropic's revenue run rate went from $1B at end of 2024 to $30B by April 2026, adding $4B in January, $7B in February, and $11B in March, with over 1,000 enterprises paying over $1M annually. 42:26
- Gerstner predicts Anthropic could exit 2026 at $80-100B in revenue, noting they are compute-constrained at 1.5-2 gigawatts and that 65-70% of enterprise token consumption is already open source. 48:18
Highlights
- 4:47
Mythos Finds Decade-Old Vulnerabilitiesdata
Anthropic's Mythos model autonomously discovered thousands of vulnerabilities including a 27-year-old bug in OpenBSD and a 16-year-old FFmpeg bug missed after 5 million automated scans.
- 6:16
Self-Regulation Over Government Mandateinsight
Gerstner argues Anthropic chose to sandbox Mythos and form Project Glasswing with 40 companies rather than calling for a government moratorium, showing market forces can handle safety without top-down regulation.
- 9:51
Anthropic's Scare Tactic Patterndebate
Sacks points out Anthropic has a proven pattern of pairing model releases with worst-case studies, citing the blackmail study where they prompted the model 200 times to get the desired result, and notes no real-world blackmail has been observed in the year since.
- 15:04
Chamath Compares Mythos to GPT-2debate
Chamath argues the Mythos rollout mirrors GPT-2's 2019 methodical release, which was hyped as the end of days but turned out to be a nothing burger. He says an advanced hacker can probably do the same things today with Opus.
- 24:04
Anthropic Ankles OpenClawstory
Jason explains how Anthropic cut off OpenClaw users' $200 subscriptions, forcing them to pay per-usage API rates, then released their own competing agent product within 10 days. Sacks raises antitrust questions about potential bundling and price discrimination.
- 42:26
$30B Revenue in Under Two Yearsdata
Anthropic went from $1B run rate at end of 2024 to $30B by April 2026, adding $4B in January, $7B in February, and $11B in March alone—more than Databricks plus Palantir combined in a single month.
- 48:18
Gerstner Predicts $80-100B Exitdata
Gerstner says he would not be shocked if Anthropic exits 2026 at $80-100B in revenue, noting they are compute-constrained at 1.5-2 gigawatts and that 65-70% of enterprise token consumption is already open source models.
- 1:11:08
Iran Ceasefire and Islamabad Talksstory
Trump agreed to a two-week ceasefire if Iran opens the strait, with Vance and Kushner heading to Islamabad for talks. Netanyahu interpreted the ceasefire to mean leveling Lebanon, dropping 160 bombs in 10 minutes.
Guests
About this episode
(0:00) Bestie intros: Brad Gerstner joins the show! (4:22) Anthropic blocks Mythos release for security concerns: serious or marketing stunt? (24:07) Are OpenAI and Anthropic trying to kill OpenClaw? Does Anthropic already have market dominance in AI coding? (42:20) Anthropic $30B run rate, fastest revenue ramp ever, the TAM for intelligence (58:01) Major vibe shift: Anthropic ripping, OpenAI reeling (1:10:12) Iran War: Ceasefire, Israel's influence, market impact Apply for Summit 2026: https://allin.com/events Follow Brad: https://x.com/altcap Follow the besties: https://x.com/chamath https://x.com/Jason https://x.com/DavidSacks https://x.com/friedberg Follow on X: https://x.com/theallinpod Follow on Instagram: https://www.instagram.com/theallinpod Follow on TikTok: https://www.tiktok.com/@theallinpod Follow on LinkedIn: https://www.linkedin.com/company/allinpod Intro Music Credit: https://rb.gy/tppkzl https://x.com/yung_spielburg Intro Video Credit: https://x.com/TheZachEffect Referenced in the show: https://www.youtube.com/watch?v=INGOC6-LLv0 https://openai.com/index/better-language-models https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf https://x.com/steipete/status/2040811558427648357 https://x.com/juliusai/status/2041292301234999668 https://polymarket.com/event/ipos-before-2027 https://www.google.com/finance/quote/IGV:BATS https://polymarket.com/event/anthropic-ipo-closing-market-cap-119 https://truthsocial.com/@realDonaldTrump/posts/116351998782539414 https://truthsocial.com/@realDonaldTrump/posts/116363336033995961 https://truthsocial.com/@realDonaldTrump/posts/116365796713313030 https://www.nytimes.com/2026/04/07/us/politics/trump-iran-war.html https://www.state.gov/releases/office-of-the-spokesperson/2026/03/secretary-of-state-marco-rubio-remarks-to-press-6
Transcript
16,339 words · 5 speakers
Speaker names marked are inferred from speaking time and the show’s known hosts — they are not voice-verified.
Topics
- AI Safety
- Anthropic Revenue
- AI Cybersecurity
- Open Source AI
- Iran Ceasefire
- Market Competition
- Israel Influence
- Antitrust
Page updated · transcript and summary are machine-generated