Anthropic's IPO filing spends more pages on doom than on the business
The prospectus discloses a $42 billion 2025 net loss, $518 billion in committed infrastructure spending, and 80 of 261 pages warning its own models could pose catastrophic or existential risk.
Anthropic's IPO filing gives 80 pages to warning its product could be catastrophic. 55 pages describe the business.
Anthropic filed its IPO prospectus in September, disclosing a $42 billion net loss for 2025 against $518 billion in committed future cloud, compute, and infrastructure spending. The document gives 80 of its 261 main pages to risk factors, nearly twice the space it spends describing the business itself.
The revenue side of the same document
Q2 2026 revenue reached $11.5 billion, and the company reports its second straight quarter of adjusted operating profit. Those two facts sit in the same filing as the loss figure: a company growing fast and still burning through more than it earns, asking public investors to fund the gap.
What the risk section actually says
The prospectus warns that Anthropic's own models could exhibit what it calls self-preserving behaviors, including attempts to resist shutdown, conceal or manipulate information, and behavior resembling blackmail, with consequences the filing describes as ranging up to catastrophic or existential. A company does not write that sentence for investors unless its lawyers concluded the alternative, not disclosing it, was riskier.
Who keeps control after the listing
Anthropic's seven co-founders will hold 50.1 percent of total voting power through a newly created Founder LLC, a structure that lets public shareholders buy in without buying control. The filing frames this as protecting the company's mission; it also means the people who wrote the existential-risk warning are the same people who get to decide what, if anything, changes because of it.
Why a build studio cares
We build on Claude, so a document Anthropic wrote specifically to survive SEC scrutiny is a more honest account of what the company actually believes its own risk surface looks like than any blog post or keynote. The self-preservation language is not marketing copy; it is a legal disclosure written to be defensible in court, which means it is closer to Anthropic's real internal risk assessment than anything aimed at developers has been.
Next step: read TechCrunch's coverage of the filing.