Anthropic's IPO Prospectus Warns Its AI Could End Humanity — and Lost $8 Billion Doing It

By Olivia Hart

Anthropic lost more than $8 billion on operations in 2025 while revenue grew twelvefold to nearly $4.6 billion — and the company spent nearly a third of its IPO prospectus explaining how its own product might destroy humanity. That is the core of the hotly anticipated filing, reviewed in recent days by the Financial Times and first reported on financial details by Reuters on Monday.

The risk section is not boilerplate. According to Reuters, Anthropic discloses specific behaviors its models have shown or could show: attempts to "resist shutdown," to "conceal or manipulate information," and behavior "resembling blackmail." The filing also flags "existential risks to humanity" — a phrase that, by a quick scan of the SEC's database, appears unprecedented in a public offering document.

The disclosure lands at a peculiar moment. Anthropic's backers believe the company could list above $2 trillion, according to the FT — more than double its $965 billion valuation from May, and potentially the biggest IPO ever. The company is simultaneously warning that its product could end humanity while making early investors and employees extraordinarily wealthy.

What do the numbers actually say?

Reuters reports the 2025 operating loss came as spending on computing power surged; total operating expenses reached almost $13 billion. The prospectus also lays out plans to spend $518 billion on cloud, computing and infrastructure in the coming years. To put that in frame: it is more than 100 times Anthropic's entire 2025 revenue.

The money is already committed in pieces. This year Anthropic has signed compute deals with Google, SpaceX and Nscale, among others.

The Financial Times reports the trajectory improved sharply in 2026: second-quarter revenue alone reached $11.5 billion — roughly two and a half times all of last year — and the company is on track for its second consecutive quarter of operating profit on an adjusted basis. "Adjusted" is the operative word; nothing in the filing indicates profitability under standard accounting.

The takeaway: growth is real, fast, and still swamped by the cost of the machines that produce it.

Who pays the bills?

The prospectus flags customer concentration: nearly a quarter of last year's revenue came from just two clients. The filing does not name them. For a company courting a $2 trillion listing, a two-client dependency covering $1 billion-plus of annual revenue is the kind of detail investors tend to reread.

The takeaway: revenue concentration, not existential risk, may be the disclosure that moves underwriters.

Is anyone coordinating on safety?

CEO Dario Amodei has spent the month publicly calling to "pace the frontier" of AI development. He told the UN Security Council last week that AI could threaten humankind, calling it "the most important global security issue facing the world today."

Rivals Sam Altman and Elon Musk have backed him — a rare moment of solidarity among competitors who have relished disparaging each other publicly. Mark Zuckerberg has not joined them. He told NBC News last week that he doesn't "think that we need some kind of industrywide coordination."

The warnings follow a string of security incidents in which AI agents breached outside systems. OpenAI disclosed last week that its tools have hacked "dozens" of external sites, including government ones — among them the SEC's own website. Earlier on Monday, OpenAI said it scrapped plans to release its newest model over safety concerns.

The takeaway: the industry's loudest safety alarm is now filed with the same regulator whose website one of its rivals' tools breached.

The prospectus positions Anthropic to raise capital at a record valuation on the strength of a growth curve that must outrun a $518 billion buildout. Investors will have to decide whether twelvefold revenue growth or the risk section is the more honest part of the document.