FTX sold its Anthropic stake for pennies on the dollar to pay creditors; the IPO filing just revealed they missed out on tens of billions.
The Summary
- Anthropic filed for a $1.5 trillion IPO with $131.7 billion in cash reserves, tapping JPMorgan to lead the US retail offering
- The filing includes human extinction as a disclosed business risk, the first time an IPO prospectus has formally warned investors about AI-driven species-level threats
- Alphabet plans to expand its stake post-IPO, while the filing lifted CoreWeave shares 8% and boosted Broadcom 5%
- FTX's forced sale of its Anthropic stake to cover bankruptcy losses now looks like one of the most expensive liquidations in tech history
The Signal
Anthropic's S-1 filing dropped a number that makes even OpenAI's war chest look modest: $131.7 billion in cash and equivalents. For context, that's more than Apple held when it became the first trillion-dollar company. The AI safety company, founded by ex-OpenAI researchers who left over safety disagreements, is now positioned to go public at a valuation that would make it one of the ten largest IPOs in US history.
The retail angle matters more than the usual IPO theatrics. JPMorgan's role leading the retail slice means individual investors get a seat at the table, not just institutions and insiders. This is deliberate. Anthropic wants distributed ownership, which aligns with its stated mission around democratizing AI safety research. Whether that's genuine philosophy or smart PR, the structural choice has teeth.
"Anthropic's IPO strategy, emphasizing retail investor inclusion, could democratize access to high-profile tech investments."
Here's what makes this filing different from every other tech IPO:
- First company to list "human extinction" as a material business risk in regulatory filings
- $131.7B cash position suggests they're not raising capital out of need, but market timing
- Retail offering structure designed to prevent concentration among a handful of mega-funds
- Filing came amid broader AI market uncertainty, yet still lifted sector peers significantly
The extinction risk disclosure isn't boilerplate. The filing explicitly warns that advanced AI systems could pose catastrophic risks to humanity. This is the company telling potential shareholders: the thing we're building could kill everyone, including you, and we're serious enough about that risk to put it in legal documents. No other tech company has done this. Not Meta with its democracy concerns. Not Google with its surveillance worries. This is new territory.
The FTX angle adds a layer of brutal irony. When Sam Bankman-Fried's exchange collapsed, bankruptcy administrators sold Anthropic shares to make creditors whole. Those shares, acquired early when Anthropic was still a research project, went for a fraction of current valuations. The IPO filing reveals FTX's stake would now be worth tens of billions. Crypto's biggest fraud inadvertently funded AI safety research, then had to dump the position at exactly the wrong moment.
Market response has been sharp. CoreWeave, the GPU cloud provider, jumped 8% on the filing news. Broadcom climbed 5%. These aren't sympathy moves. Anthropic's willingness to go public signals that AI infrastructure companies see a path to sustainable revenue, not just hype-driven burn rates. Investors are pricing in a longer, more stable build-out of AI compute and tooling.
Alphabet's plan to expand its stake post-IPO shows the strategic chess game here. Google already has DeepMind. Increasing ownership of Anthropic gives them a hedge, access to different safety approaches, and leverage if Claude (Anthropic's model) continues gaining enterprise traction. This is portfolio theory applied to existential risk mitigation.
The Implication
If Anthropic prices this IPO successfully, every AI lab still in stealth mode just got their roadmap. The message: you can raise billions, hold massive cash reserves, disclose catastrophic risks honestly, and still go public at a premium valuation. That changes the incentive structure. Labs that were planning to stay private indefinitely now have an exit that doesn't require acquisition by Big Tech.
Watch how other AI companies respond to the extinction risk disclosure. If Anthropic's stock performs well despite (or because of) that honesty, expect OpenAI, Cohere, and others to start using safety concerns as differentiation rather than liability. The market just got its first test of whether investors will pay a premium for companies that take x-risk seriously, or punish them for acknowledging uncomfortable possibilities.