Alphabet pledges up to $40B to Anthropic; Claude on Google Cloud

Alphabet announced on Friday it will invest up to $40 billion in Anthropic, committing $10 billion in cash now at a $350 billion valuation and up to $30 billion more if Anthropic meets agreed milestones. The company also pledged roughly 5 gigawatts of TPU compute and a five-year arrangement to route Anthropic’s Claude API traffic through Google Cloud.

Under the agreement, requests to the Claude API can be processed on Google Cloud infrastructure. Routing that traffic to Google’s systems will generate cloud revenue when customers run Claude-based products on Google’s network rather than on Anthropic’s own compute.

The $30 billion portion of the package is contingent on Anthropic reaching performance or usage targets. Alphabet and Anthropic have not disclosed the specific milestones or the conditions that trigger those payments.

Alphabet’s compute commitment includes dedicated TPU capacity tied to Google’s stack for five years. Alphabet said the level of capacity is intended to support companies already embedding Claude into commercial products, helping reduce the risk of capacity shortfalls or financial strain for those customers.

Several current and former employees reported internal concern at Google about allocating large compute resources to a competing model. Those employees said staff were instructed that the company must make sure Google wins “in any scenario.” Gemini and DeepMind teams continue to request GPU and TPU resources for model development and deployment.

Alphabet CEO Sundar Pichai wrote that AI use inside Google has increased, adding that “Today, 75 percent of all new code at Google is now AI-generated and approved by engineers, up from 50 percent last fall,” and that some engineering tasks have completed faster with agent-driven workflows.

The announcement comes as the U.S. Department of Defense reviews AI vendors and considers additional partnerships. A Pentagon official told reporters, “The Pentagon will continue to rapidly deploy frontier AI capabilities to the warfighter through strong industry partnerships across all classification levels.” The department is testing multiple AI platforms and evaluating rules for their use in classified and other lawful missions.

Anthropic’s Claude competes with Google’s Gemini and other large language models. The investment and compute deal create a longer-term infrastructure relationship between Alphabet and Anthropic while Anthropic remains an independent model developer.