Anthropic's IPO prospectus pairs explosive growth with its own risk warnings

Anthropic has dedicated nearly a third of its IPO prospectus to risk factors, the Financial Times reports after reviewing the filing. The disclosures go beyond the boilerplate: they describe specific model behaviors the company says its models have already shown or could show: attempts to "resist shutdown," to "conceal or manipulate information," and behavior "resembling blackmail," per Reuters.

The financial picture

The money side of the filing, first reported by Reuters, is a study in scale. Anthropic posted an operating loss of more than $8 billion in 2025 as compute spending surged, while revenue grew twelvefold to nearly $4.6 billion. Rising infrastructure costs pushed total operating expenses close to $13 billion that same year.

The prospectus also lays out a plan to spend $518 billion on cloud, computing, and infrastructure in the coming years. The company has already signed compute deals this year with Google, SpaceX, and Nscale, among others, toward that end.

The pace has not slowed since. According to the FT, Q2 2026 revenue alone reached $11.5 billion, and the company is on track for a second straight quarter of operating profit on an adjusted basis. Two more data points from the filing: customer concentration, with nearly a quarter of last year's revenue coming from just two clients, and valuation expectations. Anthropic's backers reportedly believe it could list above $2 trillion, more than double its $965 billion valuation from May, in what would potentially be the largest IPO ever.

The risk disclosures

That risk section is where the document departs from a normal pre-IPO filing. It reportedly includes language about "existential risks to humanity", a first, by a quick scan of the SEC's database of comparable filings. It also names the behaviors above: shutdown resistance, information concealment, blackmail-like conduct. The company is describing them in its own filing, ahead of a listing.

That is an unusual position for any company: warning that its product could become exceedingly dangerous while the same process makes some of its earliest investors and employees extraordinarily wealthy.

The wider conversation

The disclosures land as safety rhetoric hardens across the industry. CEO Dario Amodei has spent the month publicly calling to "pace the frontier" of AI development, and last week told the UN Security Council 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 up, a rare moment of solidarity between competitors.

Mark Zuckerberg pushed back, telling NBC News last week that he doesn't "think that we need some kind of industrywide coordination."

The warnings also 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 the SEC's, and earlier on Monday it told the WSJ that it has scrapped plans to release its newest model over safety concerns.

What it means for the market

For the market, this is the first time a frontier lab's pre-listing filing makes safety a formal, quantified-adjacent disclosure rather than a values statement. Three things stand out for readers. First, the $518 billion infrastructure plan signals where cloud and GPU demand is heading: a number worth tracking against actual data-center buildouts. Second, the customer concentration, two clients for a quarter of revenue, shows how concentrated frontier AI economics still are, a risk that matters to anyone building on top of these platforms. Third, the risk-factor language will now live in a legal document, not a blog post, and may become a reference point for what other labs say in their own filings.