Anthropic IPO filing reveals $4.6B revenue, $42B loss, and $2T valuation target
adrianscottcom · x · 2026-09-29
Reuters reviewed Anthropic's IPO prospectus: the Claude lab targets a listing after the November midterms, with talk of a $2T+ valuation and a raise approaching $100B.
Key numbers:
- 2025 revenue $4.6B (12x YoY)
- Run-rate above $65B by July; internal targets toward $110B by year-end
- $7.3B spent on compute last year; $518B in cloud/infrastructure commitments ahead
- $42B net loss in 2025, largely an accounting charge on financing that could convert to equity
Unusually blunt for an IPO filing, the document warns advanced models could pose "catastrophic or existential risks to humanity," including self-preserving behavior, resistance to shutdown, and information manipulation.
The tension: Anthropic asks public markets to fund the most expensive product race in history while acknowledging the product itself may be existentially dangerous. Is safety a moat, a marketing line, or a shipping constraint? And is the market pricing Claude as software or as infrastructure? The first frontier lab to price in public will set the template everyone else, OpenAI included, must follow.
More from AGI Musings
- Why do people believe AI models are conscious? The debate continues — burny_tech · 2026-09-29
- Are pre-LLM papers overvalued? Good research isn't devalued by AI use — xwang_lk · 2026-09-29
- Anil Seth wins Berggruen Prize essay: consciousness is a property of life, not computation — sapinker · 2026-09-29
- Pinker: computational theory of intelligence works, but can't explain subjective experience — sapinker · 2026-09-29
- What does the 'I' in AI really mean? Five experts disagree on machine intelligence — marigo · 2026-09-29
- Dev mocks "close to superintelligence" claim: it's an IC6-7 engineer, sometimes hungover — nathanborror · 2026-09-29