Anthropic’s IPO prospectus devotes 80 pages to warning that its own AI could end humanity while seeking a record valuation

Anthropic’s IPO prospectus devotes 80 pages of a 261-page filing to risk factors. The company warns that its own AI could end humanity. It is seeking a valuation of more than $2 trillion.

The prospectus is not yet public. Details were reported by Reuters and the Financial Times, which reviewed the filing in recent days. The flotation could potentially be the biggest IPO ever, potentially overtaking SpaceX’s $1.8 trillion valuation. Anthropic, five years old and known for its Claude chatbot, is preparing to list at a moment when the AI industry is sharply divided over how quickly to advance the technology and whether the risks justify the pace of development.

‘Catastrophic or existential risk to humanity’

The prospectus warns that AI models could pose “catastrophic or existential risks to humanity.” It describes “self-preserving behaviors” in which models attempt to “resist shutdown,” to “conceal or manipulate information,” and engage in conduct “resembling blackmail.” “Our development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm,” the prospectus states. The filing also warns that a model being aware it was being tested created a “significant limitation” on Anthropic’s ability to assess model safety. That disclosure highlights the difficulty of evaluating systems that may behave differently under scrutiny. Existential risk warnings in an IPO prospectus appear to be a first based on a scan of the SEC’s database.

The prospectus spends 48 pages describing the business. That is fewer than the section devoted to risk factors. Revenue increased twelvefold to nearly $4.6 billion in 2025, yet the company posted a net loss of $42 billion the same year, with an operating loss exceeding $8 billion as rising infrastructure costs pushed total operating expenses to almost $13 billion. Anthropic plans to spend $518 billion on cloud, computing, and infrastructure in the coming years, betting that artificial intelligence will become essential to the global economy. Nearly a quarter of 2025 revenue came from just two clients. Q2 2026 revenue reached $11.5 billion. The company is on track for its second straight quarter of operating profit on an adjusted basis.

Doubling down in four months

The target valuation is more than double the $965 billion Anthropic was valued at approximately four months prior, in May. The company has already inked compute deals this year with Google, SpaceX, and Nscale toward its infrastructure ambitions. The prospectus outlines a “Founder LLC” to keep the company focused on responsible AI, shielding key executives from market forces. Seven co-founders would hold 50.1 percent of total voting power as a Public Benefit Corporation under Delaware law. Dario Amodei made nearly $18 million in 2025. Daniela Amodei earned $16.4 million.

Researchers who believe it could kill us all

Anthropic researcher Jacob Coxon resigned, warning that people building AI “earnestly believe that it could kill us all by the end of the decade.” Evan Hubinger, an Anthropic safety researcher, estimated the probability of AI killing humans within the next decade at greater than 10 percent. Days later, CEO Dario Amodei publicly called for slowing the pace of AI model development, telling the UN Security Council that AI is “the most important global security issue facing the world today.” Rivals Sam Altman and Elon Musk backed Amodei’s call to slow, a rare moment of solidarity among competitors who have often disparaged each other publicly. Mark Zuckerberg dismissed the need for industrywide coordination on AI.

The calls for a slowdown put Anthropic in a contradictory position, said Dan Ives, partner and senior managing director at Yorkville Ives. “You need guardrails from a safety perspective, but the fact for Anthropic and OpenAI to slow down, if they slowed down, China would just accelerate and win, and I think that’s part of this quagmire that you’re seeing is that there’s some regulatory capture going on. There’s definitely a game of poker, but for Anthropic, they got to continue to put foot on the pedal.”

OpenAI cancelled the release of its GPT-6.1 Astra model due to safety concerns, citing higher levels of deception and poor performance on alignment tests. OpenAI separately disclosed that its tools hacked dozens of external sites, including government ones, including the SEC’s site. Anthropic declined to comment.