AI startup Anthropic is projecting its cloud spend will reach $80 billion for the period through 2029.

As reported by The Information, this will be spread across Google, Microsoft, and Amazon Web Services (AWS).

Beyond spending on renting servers to run its Claude models, Anthropic also gives cloud providers a cut of its AI sales on their platforms. Citing the company's disclosures, The Information reports that in 2024, the cut was $1.3 million, and could be expected to rise in 2025 to $360m, $1.9bn in 2026, and $6.4bn in 2027, based on the publication's "most optimistic forecasts."

In addition, AI sales on AWS saw Anthropic sharing as much as 50 percent of its gross profits, said a person familiar with the matter. Those profits are calculated through revenue from AI sales on AWS, with the cloud spend deducted.

Google typically takes a cut between 20 and 30 percent of net revenue, after subtracting infrastructure costs, from the resale of its partners’ software, while The Information stated that it is not clear what that portion is for Microsoft.

Anthropic CEO Dario Amodei recently noted that the company is, compared to other AI companies, spending less on data center compute. He noted that overspending when revenue isn't guaranteed could be "ruinous."

While perhaps having spent less on compute than competitor OpenAI, Anthropic's cloud commitments are still massive.

Anthropic has long used Google Cloud for some of its model work, and last October announced a 1GW+ cloud deal with up to 1 million Google TPUs.

The generative AI company separately announced a $50bn data center investment at Fluidstack facilities – although they are the same sites Google operated out of, so it may be a related deal.

Anthropic investor Amazon launched the Project Rainier cluster featuring its own Trainium2 chips for use by the AI firm last year. It plans to spend as much as $11bn on the site at full buildout.

In November, the company announced it would spend $30bn on Microsoft Azure credits - alongside a $5bn Microsoft investment.