Anthropic Cloud Crunch: Over $750 Billion Compute Bill Shadows November IPO
Anthropic may spend over $750 billion on cloud capacity as an $80.7 billion loss, supplier tensions and AI liability risks shadow its planned November IPO.
Summary
A leaked S-1 filing available by October 20, 2026, says Anthropic could spend more than $750 billion on cloud contracts, including deals still under negotiation, yet cannot secure enough capacity. AWS, Google and Microsoft increasingly view the Claude maker as a competitor rather than a customer. Anthropic is leasing its own data centers and working with third-party operators, but warns their limited operating histories could make them less reliable than hyperscalers.
Anthropic recorded an $80.7 billion net loss against $16.3 billion in revenue during the first half of 2026, while two customers generated about one-fifth of sales. The filing warns that liability for harmful AI behavior could exceed available insurance coverage. Anthropic told prospective investors last week that third-quarter revenue reached $16.8 billion, 45% above the previous quarter, and is targeting a November IPO. The company did not immediately comment on the filing.
Positives
- Third-quarter revenue reached $16.8 billion, rising 45% from the previous quarter.
- First-half revenue totaled $16.3 billion despite Anthropic’s heavy infrastructure spending.
- Anthropic is adding capacity by leasing data centers and working with third-party operators.
- A November IPO could give Anthropic access to public capital as its cloud requirements soar.
Risks & concerns
- More than $750 billion in potential cloud contracts may still leave Anthropic short of capacity.
- An $80.7 billion first-half net loss sharply exceeded Anthropic’s $16.3 billion revenue.
- Two customers generated about one-fifth of sales, creating substantial revenue concentration.
- AWS, Google and Microsoft increasingly regard Anthropic as a competitor, complicating critical supplier relationships.
- Third-party data center operators have limited operating histories and may be less reliable than hyperscalers.
- Damages from harmful AI behavior could exceed Anthropic’s available insurance coverage.