Leaked S-1 shows Anthropic struggling for cloud capacity as rivals eye it warily

A leaked S-1 filing, reported by The Wall Street Journal, shows Anthropic can’t line up all the cloud capacity it needs from AWS, Google, and Microsoft — hyperscalers that increasingly treat the Claude maker as a rival rather than a client. The filing adds detail to figures Reuters and the Financial Times reported in late September: Anthropic believes it could spend more than $750 billion on cloud contracts, some still under negotiation.

Even that isn’t enough capacity. Anthropic is leasing its own data centers and working with third-party operators, though it worries those operators’ “limited operating history” could make them less dependable than the major cloud providers.

The spending follows a first-half net loss of $80.7 billion; Anthropic’s revenue over that same stretch was $16.3 billion, with just two customers responsible for about a fifth of sales, per the WSJ report. The filing also warns, building on previously disclosed catastrophic-risk language, that Anthropic could face liability for AI-related harm that exceeds its available insurance coverage.

Anthropic told prospective investors last week that third-quarter revenue alone reached $16.8 billion, up 45% from the prior quarter. The company is reportedly aiming for an IPO in November. Anthropic did not immediately respond to a request for comment.

Why it matters

A company burning cash at this scale while being frozen out of the cloud capacity it needs — by the same hyperscalers it’s trying to out-compete — shows how AI’s infrastructure bottleneck is now also a competitive one.