According to leaked S-1 documents seen by The Wall Street Journal, Anthropic has struggled to secure sufficient cloud capacity from AWS, Google, and Microsoft, as these cloud providers increasingly view the Claude developer as a competitor rather than a customer.

The leaked documents provided more details on the figures reported by Reuters and The Financial Times at the end of September, including Anthropic’s belief that its cloud contract expenditures could exceed $750 billion, with some deals still under negotiation.

Even so, it’s still not enough. Anthropic is leasing its own data centers and collaborating with third-party data center operators. However, the company is concerned that these operators have “limited operational history,” which may make them less reliable than hyperscale enterprises.

Behind Anthropic’s massive spending lies a net loss of $80.7 billion in the first half of this year. According to The Wall Street Journal, the company’s revenue during the same period was $16.3 billion, with just two clients accounting for about one-fifth of its sales.

Building on the previously reported disclosure of catastrophic risks, the document also warns that Anthropic may be held liable for damages caused by harmful AI behavior, potentially exceeding available insurance coverage.

Last week, the company informed potential investors that its revenue for the third quarter alone reached $16.8 billion, a 45% increase from the previous quarter. It is reported that Anthropic aims to go public in November.