Anthropic’s leaked IPO filing pairs AI doom warnings with surging costs

A leaked draft of Anthropic’s IPO prospectus pairs rapid revenue growth and steep losses with warnings that its AI could endanger humanity, Reuters reported on September 28.

The disclosures, covered in The Rundown’s September 30 briefing, follow Anthropic’s June 1 announcement of a confidential IPO filing. Anthropic said the filing gave it the option to go public after SEC review, subject to market conditions and other factors.

The cost of growth

Anthropic recorded about $4.6 billion in revenue in 2025, roughly 12 times its 2024 total, while its operating loss exceeded $8 billion, TechCrunch reported, citing Reuters and the Financial Times. Risk factors take up nearly a third of the prospectus, according to the report.

The company has also committed at least $518 billion to infrastructure over the next decade, according to TechTarget’s account of Reuters’ reporting. About 80% of those commitments reportedly cannot be canceled or must be paid regardless of how much capacity Anthropic consumes.

Anthropic’s more recent figures describe a business that has grown substantially since 2025. In its May 28 financing announcement, the company said it had raised $65 billion, valuing it at $965 billion after the investment. It also said its revenue run rate had crossed $47 billion earlier that month, an annualized measure of sales at that point.

Why it matters

Saturday Night Live had just mocked Dario Amodei, with Jane Wickline playing an executive asking lawmakers to restrain his own activity. A prospectus that devotes so much space to danger feels like an extension of the sketch. Anthropic is preparing to seek public capital for the same technology it warns could endanger humanity.

On September 12, Axios reported that safety concerns had delayed IPO plans at Sam Altman’s OpenAI. Anthropic, so vocal about AI danger, appears to be pressing ahead while its rival slows its plans. Which company will list first remains uncertain.

The spending commitments make the tension concrete. If Anthropic must pay for capacity even when it sits idle, slowing releases for safety reasons could reduce revenue growth without cutting costs by the same amount. That could put pressure on leaders to keep deploying products when caution calls for a pause.

New capital can also fund safety work. Anthropic said its May financing would support safety research, computing capacity and product growth. The hard question is how management would balance those goals if a safety finding called for a costly delay.

An eventual public prospectus should clarify who can halt a release, how leaders settle safety disputes, and whether research budgets have protection during a slowdown. Those decisions could affect whether researchers’ concerns lead to action and when customers receive new Claude capabilities. Prospective shareholders need to understand what a pause would cost while infrastructure payments keep coming.