Anthropic has committed to spending roughly $200 billion on Google Cloud services and custom chips over the next five years, according to a report from The Information picked up by Reuters, CNBC, Engadget and other outlets late on May 5. The commitment, if confirmed, would rank among the largest cloud spending agreements ever disclosed and underscores how concentrated frontier AI economics have become around a handful of compute providers.
Alphabet shares rose about 2% in extended trading on Tuesday after the report. Anthropic and Google did not respond to requests for comment.
Forty Percent of Google's Backlog
The most striking number is not the headline dollar figure but the share of Google's pipeline it represents. Per The Information, Anthropic alone accounts for more than 40% of the revenue backlog Alphabet disclosed to investors last week. Backlog reflects contractual commitments from cloud customers that have not yet been recognized as revenue, and is a key metric investors use to gauge future cloud growth.
Reuters reported that contracts involving Anthropic and OpenAI now make up more than half of the roughly $2 trillion in backlogs at major cloud providers. That concentration helps explain why hyperscaler stocks have moved so closely with frontier-lab funding announcements over the past quarter.
Tied to the Broadcom TPU Deal
The $200 billion commitment builds on a deal Anthropic signed in April with Google and chip partner Broadcom for multiple gigawatts of tensor processing unit capacity, with capacity expected to start coming online in 2027. That earlier announcement was framed as a multi-year buildout but did not put a precise dollar value on Anthropic's side of the commitment.
The new figure puts Anthropic's compute spend on roughly the same trajectory as OpenAI's previously reported infrastructure deals with Microsoft, Oracle, AWS and CoreWeave, signaling that the company's compute requirements have scaled to match its closest rival despite a smaller revenue base.
Implications for the AI Compute Market
For Google, the report reinforces the strategic importance of its in-house TPU stack and its Broadcom partnership as it competes with NVIDIA-centric hyperscalers. Locking in Anthropic as a multi-year anchor tenant gives the cloud unit a forecastable revenue stream against which to justify continued capex.
For Anthropic, the commitment is a double-edged sword. It secures the gigawatt-scale capacity needed to train successor models to Claude Opus 4.7 and to serve enterprise customers like Moody's, NEC and the Blackstone, Hellman & Friedman, and Goldman Sachs joint venture announced earlier this month. But it also locks the company into payment obligations that will require either rapid revenue growth or sustained primary fundraising — Anthropic is reportedly in talks to raise at a roughly $900 billion primary valuation and trades at over $1 trillion on secondary markets.
The broader signal: the era of AI labs renting incremental capacity is over. Frontier compute is now negotiated in nine- and ten-figure multi-year blocks, and the gap between labs that have signed those blocks and labs that have not is widening.
