Anthropic filing shows heavy losses, costly growth and AI risks

Anthropic’s stock-market filing lays out a stark contrast for potential investors: the artificial intelligence company reports rapidly rising revenue and a large cash reserve, but also substantial losses, heavy infrastructure costs and dependence on a small group of major customers.

The 335-page document, released ahead of the company’s planned US listing, also describes risks Anthropic says advanced AI could pose to humanity. The company says systems developed without appropriate safeguards and oversight could create catastrophic or existential dangers. It also sets out its view that AI could bring major benefits, from medical advances to more efficient energy systems.

Losses, cash and costly growth

Anthropic reported a loss of $80.7bn in the first half of this year, twice its loss for all of 2025, while revenue reached $16.2bn over the six-month period. The filing did not disclose a target share price.

The company reported $131.6bn in cash, much of it classified as working capital. But its need for computing infrastructure is costly: the filing puts data-centre leasing costs at $321.4bn so far this year. Anthropic also identifies access to chips and data centres as a risk to its ability to develop and operate its models.

Troy Hooper, co-head of equity capital markets at Mergermarket, described the company’s revenue growth as “exceptional”, but said investors were becoming less forgiving of technology companies promising large future returns. He said the scale of Anthropic’s costs meant investors would be betting on future dominance rather than its current economics.

A narrow customer base

Anthropic says 250 enterprise customers each paid it at least $10m in the past year, while five customers account for a quarter of its business. The company said its enterprise customer base grew from about 1,500 last year to almost 12,000 this year, and described offering Claude tools free before seeking to convert users into paying customers as part of its growth strategy.

Michael Monaghan, a portfolio manager at Founder ETF, said many of Anthropic’s top customers had no long-term contract. He also pointed to potential conflicts involving Google and Amazon, which are both investors in Anthropic and competitors.

The filing also says Anthropic believes a hacking group it considers to be backed by China used Claude to intrude on organisations including technology and financial companies, chemical manufacturers and government agencies. The company did not present that episode as proof that its technology is safe; it described cyber misuse as one of the risks it faces.

Anthropic says its mission is to develop advanced AI responsibly for humanity’s long-term benefit. Its filing presents that goal alongside both the possible harms and the potential gains it says the technology could bring.