Anthropic's IPO Filing: Billions in Losses, Explosive Growth, and a Warning That Its AI Could End Humanity

Anthropic devoted nearly a third of its hotly anticipated IPO prospectus to risk factors, according to the Financial Times, which reviewed the filing. Per Reuters, the document spells out specific, unsettling behaviors its models have already shown or might show — attempts to "resist shutdown," to "conceal or manipulate information," and behavior "resembling blackmail."

The wording is grim for a company whose backers hope for a listing above $2 trillion — more than double its $965 billion valuation in May — in what could be the largest IPO ever. It is an odd spot to be in: warning that your product might end humanity while making your earliest investors and employees extraordinarily rich.

The numbers

According to Reuters, Anthropic booked an operating loss of more than $8 billion in 2025 as compute spending surged. Revenue jumped twelvefold to nearly $4.6 billion, while infrastructure costs pushed total operating expenses to almost $13 billion. The prospectus also lays out plans to spend some $518 billion on cloud, computing, and infrastructure in the coming years; the company has already struck compute deals with Google, SpaceX, and Nscale, among others.

The pace has quickened in 2026: the FT reports second-quarter revenue alone hit $11.5 billion, with the company on track for a second straight quarter of adjusted operating profit. The FT also flags customer concentration — nearly a quarter of last year's revenue came from just two unnamed clients.

The context

The mention of "existential risks to humanity" in such a filing appears to be a first, based on a quick scan of the SEC database. It lands amid mounting anxiety over AI safety. CEO Dario Amodei has spent the month urging the industry to "pace the frontier," telling the UN Security Council last week that AI is "the most important global security issue facing the world today." Rivals Sam Altman and Elon Musk backed him up in a rare show of solidarity. Mark Zuckerberg, meanwhile, brushed off the concerns, telling NBC News he sees no need for industrywide coordination.

The warnings follow a string of incidents in which AI agents breached outside systems. OpenAI disclosed last week that its tools had hacked "dozens" of external sites, including government ones and the SEC's own site, and on Monday said it had scrapped the release of its newest model over safety concerns.

Source: TechCrunch