Google plans to invest up to $40 billion in Anthropic, Bloomberg reported, with an initial $10 billion commitment at a $350 billion valuation and a further $30 billion contingent on performance targets. The deal expands an existing infrastructure relationship, with Google Cloud now providing Anthropic with a fresh five gigawatts of computing capacity over five years, adding to a previously announced arrangement with Google and chipmaker Broadcom for 3.5 gigawatts of TPU-based capacity from 2027.
The investment arrives as Anthropic faces mounting pressure on multiple fronts. The company has fielded widespread complaints about Claude usage limits and responded with a string of infrastructure deals, including a $5 billion investment from Amazon tied to a commitment to spend up to $100 billion across AWS, and a separate data centre agreement with CoreWeave. Investors have reportedly been seeking to back Anthropic at valuations of $800 billion or more, and the company is said to be considering an IPO as early as October.
Separately, Anthropic disclosed results from Project Deal, an internal experiment in which AI agents represented 69 employees as buyers and sellers in a classified marketplace. Participants were given $100 budgets, and agents completed 186 transactions totalling over $4,000. Anthropic found that users represented by more capable models achieved measurably better outcomes — but that those on the losing side typically failed to notice the disparity, raising concerns about invisible quality gaps in agent-mediated commerce.
