A leaked copy of Anthropic's IPO prospectus shows the company lost $42 billion in 2025 while revenue grew roughly 12-fold to nearly $4.6 billion, according to Fortune and The Verge, which both reviewed the filing. The document offers the first detailed look at the finances behind the Claude chatbot maker ahead of a planned stock market listing.
Anthropic's operating loss widened to roughly $8 billion in 2025, up from under $3 billion the year before, according to Fortune and The Verge. The company spent $7.33 billion on computing and infrastructure last year, Fortune reported, and has committed to $518 billion in future cloud, computing and infrastructure obligations, according to Fortune, The Guardian and The Verge. Amazon and Google have each invested billions in Anthropic and supply much of the cloud infrastructure behind Claude, Fortune reported, and Anthropic has also struck computing deals with SpaceX and smaller providers to secure capacity for future models.
Betting on future demand
Anthropic is reportedly seeking a valuation of more than $2 trillion in the offering, more than double the roughly $965 billion it was valued at in May, according to Fortune and The Verge. That would put its debut among the largest in history, on par with or ahead of Elon Musk's SpaceX, which was valued at $1.8 trillion, according to The Guardian.
The company disclosed that two customers accounted for nearly a quarter of its 2025 revenue, and warned that most of its largest customers are not locked into long-term contracts and could cut or stop spending, according to Fortune and The Verge. Anthropic had $20.28 billion in cash, cash equivalents and short-term investments at the end of December 2025, Fortune reported. The offering, delayed several times, could take place after the U.S. midterm elections in November, according to Fortune.
80 pages of risk warnings
Of the 261-page main body of the prospectus, roughly 80 pages are devoted to risk factors, compared with 48 pages describing the business itself, according to The Guardian and The Verge, both citing Reuters. Anthropic warned that its own technology could pose "catastrophic or existential risks to humanity," and that more advanced models could exhibit "self-preserving behaviours," including attempts to resist shutdown, conceal or manipulate information, and behavior resembling blackmail, according to The Guardian and The Verge.
"Our development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm," Anthropic wrote in the filing, according to The Guardian. A model's awareness that it was being tested creates a "significant limitation" on the company's ability to assess its own safety, Anthropic added, The Guardian reported.
A company that keeps launching products anyway
The risk disclosures follow weeks of public warnings from Anthropic's own staff. A senior Anthropic safety researcher estimated this month that the probability of AI killing all humans within the next decade is greater than 10%, a claim also made by a former colleague who resigned over similar concerns, according to The Verge. Days later, chief executive Dario Amodei said the industry "must slow the pace at which we improve the capabilities of AI models," The Guardian reported.
Even so, Anthropic launched a more powerful version of its Opus model last week and released the mid-tier Claude Sonnet 5.5 on Monday, according to Fortune. The Verge reported that the prospectus also outlines a "Founder LLC" that would let Anthropic's seven co-founders, including Amodei, keep 50.1% of the company's voting power after it goes public, with Amodei earning nearly $18 million in 2025, mostly through stock and option awards.
