Anthropic's S-1 details $8.06B operating loss, $518B in commitments, and existential-risk warnings ahead of a $2T-targeted IPO

Anthropic's IPO filing shows twelvefold 2025 revenue growth to nearly $4.6 billion against an $8.06 billion operating loss and $518 billion in planned infrastructure commitments, while devoting about a third of the document to risks including that its models…

Anthropic has filed an S-1 ahead of a public debut The Decoder says is likely in November, with the prospectus reviewed by Reuters and the Financial Times. About a third of the document — roughly 80 pages, per The Verge — is devoted to risk factors warning that advanced Claude models could resist shutdown, conceal or manipulate information, or exhibit blackmail-like behavior, and could cause existential harm; The Verge characterizes the risk as 'catastrophic,' wording the other reports do not use. On finances, The Decoder reports the 2025 operating loss widened from $2.98 billion to $8.06 billion as revenue grew twelvefold to nearly $4.6 billion; Reuters and Ars Technica describe the loss as more than $8 billion on operating expenses near $13 billion. The Verge separately reports a $42 billion net loss, a figure no other report gives — the discrepancy with the operating-loss number is unreconciled across outlets. Compute and infrastructure spending reached $7.33 billion in 2025, and Anthropic has $518 billion in total planned cloud, compute, and infrastructure commitments, including deals with Google, SpaceX, and Nscale. The FT reported $11.5 billion in Q2 2026 revenue, with two customers accounting for nearly a quarter of last year's revenue; Ars Technica adds the company is on course for a second consecutive quarter of adjusted operating profit. Backers reportedly see a listing above $2 trillion, versus a $965 billion valuation in May, and The Verge reports a Founder LLC including CEO Dario Amodei and six co-founders would retain 50.1 percent of voting power. For context, Ars Technica notes current and former staff have publicly warned runaway AI could end humanity within a decade, and that Amodei told the UN Security Council AI is the most important global security issue today. Separately, Ars Technica reports OpenAI scrapped plans to release its newest model next month, citing safety concerns.

• Anthropic filed an S-1; The Decoder reports the IPO debut is likely in November, with backers targeting a valuation above $2 trillion versus a $965 billion valuation in May.
• 2025 revenue grew about twelvefold to nearly $4.6 billion; The Decoder puts the 2025 operating loss at $8.06 billion (widened from $2.98 billion) on operating expenses near $13 billion.
• The Verge reports a $42 billion net loss, a figure no other report gives; others cite the operating loss instead, and the discrepancy is unreconciled.
• 2025 compute and infrastructure spending reached $7.33 billion; total planned cloud, compute, and infrastructure commitments are $518 billion, including deals with Google, SpaceX, and Nscale.
• Q2 2026 revenue was $11.5 billion (FT); Ars Technica says Anthropic is on course for a second consecutive quarter of adjusted operating profit.
• Two customers accounted for nearly a quarter of 2025 revenue.
• About a third of the prospectus — roughly 80 pages per The Verge — covers risk factors: models resisting shutdown, concealing or manipulating information, blackmail-like behavior, and existential risk; only The Verge uses the word…
• The Verge reports a Founder LLC including CEO Dario Amodei and six co-founders would retain 50.1 percent of voting power.