AI start-up Anthropic has agreed to pay Google some $200 billion (£148bn) over the next five years for access to cloud computing resources, The Information reported, suggesting that the company accounts for more than 40 percent of a cloud revenue backlog Google reported last week.
The terms are reportedly part of a deal announced in late April under which Google, a major Anthropic backer, said it would invest $10bn into Anthropic at its current $380bn valuation, a figure that could rise to $40bn depending on performance milestones.
Circular deals
The reported deal is the latest circular arrangement among large tech companies operating in the AI sector, under which major players have invested in one another and received large supply contracts in return.
Such deals have given the companies concerned a healthy stock price boost, with Google currently approaching top-ranked Nvidia in market value, in part due to soaring demand for its cloud services from AI companies.
Google reported major cloud growth and increased capital expenditure expectations in late April.
Deals with Anthropic and OpenAI currently account for a revenue backlog of $2tn across Amazon Web Services, Google Cloud and Microsoft Azure, The Information report said.
Cloud infrastructure
Under a deal earlier in April with Google and chip designer Broadcom, Anthropic gained access to multiple gigawatts of computing power via Google’s Tensor Processing Units, which it expects to come online starting in 2027.
Also in April, Anthropic struck a multi-year agreement with infrastructure company CoreWeave and is planning to secure nearly 1 gigawatt of capacity via Amazon’s AI chips by the end of this year.
Google’s Gemini is also a direct competitor to Anthropic’s Claude models.
