Anthropic adds $10 billion to cloud bill as Lambda deal wobbles
Anthropic has agreed to rent about $10 billion more in cloud computing capacity from Microsoft and Google, according to a filing with US regulators on Tuesday.
The new commitments, made over the past month, run until 2033.
They take Anthropic's minimum commitment to Microsoft to about $44.9 billion.
Its minimum commitment to Google rises to about $121.3 billion.
Anthropic, which makes the Claude chatbot, is locking in huge amounts of computing power as demand for its AI models grows.
The latest deals follow an agreement in November 2025 under which Microsoft and Nvidia agreed to invest up to $5 billion and $10 billion in Anthropic, respectively.
As part of that deal, Anthropic pledged to spend about $30 billion on Microsoft's Azure cloud service.
It has also committed at least $110 billion to Amazon for cloud capacity between May 2026 and April 2036.
Across all its suppliers, Anthropic's infrastructure commitments now total about $895.4 billion, with some running until 2045.
Lambda setback
The filing also revealed problems with Lambda, a company that rents out AI chips in the cloud.
Anthropic signed a deal with Lambda in August worth about $34.9 billion, part of a wider relationship valued at about $41.3 billion through 2034.
Lambda missed a deadline to secure the lease on a key data centre, which gives either company the right to cancel the $34.9 billion deal.
Anthropic said neither side had issued a notice to end the agreement, and it did not currently plan to cancel.
Read more about AI ethics
- 02/11/26Anthropic and numbers you need to know. Or how the Claude giant has racked up nearly $900bn of compute commitments
- 29/09/26Anthropic's leaked prospectus shows how much AI money flows back to its backers
- 06/05/26Alphabet briefly overtakes Nvidia after report Anthropic committed $200bn to Google Cloud
