Anthropic S-1 Leak: $2 Trillion IPO Valuation, $4.6B Revenue and $8B Loss
Anthropic's S-1 IPO prospectus has leaked: the company plans to go public at a $2 trillion valuation, and its financials and risk disclosures have drawn widespread scrutiny of its valuation logic and agent narrative.
Confirmed
- According to Fortune (relayed by @gerardsans and @dylan522p), the prospectus shows the company plans to IPO at a $2 trillion valuation, with losses still enormous.
- Figures disclosed by Reuters (relayed by @量子位): revenue in 2025 surged 12x to nearly $4.6 billion; operating losses doubled to over $8 billion; roughly 1/4 of revenue comes from two unnamed major customers.
- The company has committed at least $518 billion in spending for AI infrastructure over the next decade, 80% of which is non-cancellable or "take-or-pay".
- The prospectus also includes a warning that AI "poses an existential risk to humanity" (relayed by @LexSokolin).
Doubts and controversy
- SemiAnalysis founder Dylan Patel quipped about the valuation dilemma: investing at a $2 trillion valuation means either the company goes to zero and he loses everything, or it grows to $20 trillion and "we're all doomed" — reflecting how top AI companies' valuations have left investors stuck.
- @gerardsans questioned the prospectus's core narrative of "agent labor that can run for weeks," arguing today's agents are more like assistants sitting beside you that need constant supervision, far from weeks of continuous autonomous operation.
- Lex Sokolin mocked the prospectus for containing "two doomsdays in one document": an existential risk to humanity, and the $8 billion loss on the balance sheet.
Why it matters
- This is a rare large-scale exposure of a top AI lab's financials: 12x revenue growth alongside doubled losses, plus a decade-long, $518 billion infrastructure commitment that is 80% non-cancellable, sketching a business model where extreme growth is tightly bound to extreme capital consumption.
- Heavily concentrated revenue from two major customers, combined with the gap between grand agent narratives and current reality, will be a focal point for IPO pricing and subsequent regulatory and investor scrutiny.
2026-10-01 ~ 2026-10-02 · 5 related posts
- Episode 1: Anthropic Warns of AI Risks While Racing Toward Record IPO(2026-09-16, 2 posts)
- Episode 2: Anthropic Targets November IPO, Weighs New Model to Counter GPT-6 Astra(2026-09-19, 20 posts)
- Episode 3: Anthropic Files for IPO: $2 Trillion Target, $42B Net Loss(2026-09-29, 65 posts)
- Episode 4: Anthropic S-1 Leak: $2 Trillion IPO Valuation, $4.6B Revenue and $8B Loss(2026-10-01, 5 posts)
- Episode 5: Anthropic Targets Mid-November IPO at Up to $2 Trillion Valuation(2026-10-02, 4 posts)
Primary sources
- Anthropic's $2 trillion IPO prospectus leaks, revealing steep losses — gerardsans ·
- Anthropic Prospectus Reveals $4.6B Revenue, 25% From Two Mystery Clients and $518B in Compute Pledges — 量子位 ·
- Anthropic's S-1 warns of 'existential risks to humanity' while reporting an $8 billion loss — LexSokolin ·
- SemiAnalysis' Dylan Patel Can't Buy Anthropic at $2T: "It's a 0 or We're All Fucked" — dylan522p · 2026-10-01
- [source] Anthropic's S-1 warns of 'existential risks to humanity' while reporting an $8 billion loss — LexSokolin · 2026-10-01
- [source] Anthropic Prospectus Reveals $4.6B Revenue, 25% From Two Mystery Clients and $518B in Compute Pledges — 量子位 · 2026-10-01
- Anthropic Wants $2T for Week-Long Agents; Skeptic Says It's Just Excel Waiting to Be Fed — gerardsans · 2026-10-01
- [source] Anthropic's $2 trillion IPO prospectus leaks, revealing steep losses — gerardsans · 2026-10-01