Anthropic discloses documents ahead of stock market listing: losses, risks and dependence on five clients

What the company disclosed

Anthropic, the owner of the Claude chatbot, has published a 335-page package of documents required for its long-awaited debut on the US stock market, BBC News reports. The company has not yet disclosed its target offering price, but the market expects it could go public with a valuation of $2 trillion or more — which would make the listing the largest in history. The documents published by Anthropic show how the company earns money, what it spends it on and who runs it, and also describe the risks of its technology — up to and including threats to human lives.

Losses are growing, but there is plenty of money

The financial figures in the filing are stark. In the first half of 2026, Anthropic lost $80.7 billion — twice as much as in all of 2025 — on revenue of $16.2 billion over six months. Revenue is growing fast, but the company is still far from breaking even. At the same time, it has an enormous cash reserve: $131.6 billion in cash, almost all of it classified as working capital. According to BBC calculations, Anthropic currently has more money on hand than Meta and Google combined. The main expense item is infrastructure: data center rentals alone cost the company $321.4 billion since the start of the year, and the company calls access to chips and computing power one of the key factors behind its success amid competition with other AI giants for talent and compute.

Dependence on a few clients

Anthropic's business is concentrated on a narrow group of customers: 250 large corporate clients, each of which paid at least $10 million over the year, account for a significant share of revenue, and just five clients account for a quarter of the business. This makes the company dependent on a small share of paying customers. At the same time, the corporate client base is growing rapidly: from roughly 1,500 last year to nearly 12,000 this year. The growth strategy includes free access to Claude tools followed by converting users to paid plans; the company claims that Claude has achieved 'the fastest enterprise adoption of any technology in history.' The filing also notes seasonality: in January, customer spending fell by 20%, and the use of AI tools declines, in particular, during the Christmas period.

Risks — up to 'existential' ones

Among its risk factors, Anthropic stated, in a way unique for a public company, that 'sufficiently advanced AI systems, if not developed, deployed and operated with proper safeguards, oversight and safety, could pose catastrophic or existential risks to humanity.' The company cited an incident from less than a year ago: a hacking group that Anthropic believes is sponsored by China 'manipulated' Claude to break into major technology and financial companies, chemical plants and government agencies. Alongside this, the company repeats its thesis about the 'extraordinary' benefits of AI: personalized medicine and accelerated drug development, individualized education, legal and financial advice, more efficient power grids and batteries, as well as helping the developing world. The company's mission is formulated as 'responsible development and maintenance of advanced AI for the long-term benefit of humanity.'

Image source, EPA