Anthropic inks multi-gigawatt TPU deal with Google, Broadcom to scale AI compute

In a major push to scale infrastructure for its artificial intelligence (AI) models, Anthropic on Monday announced signing a new agreement with Google and Broadcom to secure multiple gigawatts of next-generation tensor processing unit (TPU) capacity.

The company said the additional compute capacity is expected to begin coming online from 2027 and will support its flagship Claude models amid surging global demand.

“This groundbreaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure,” Anthropic Chief Financial Officer Krishna Rao said, adding that the company was making its “most significant compute commitment to date.”

Anthropic said demand for its Claude AI models has accelerated in 2026 and the annualised revenue run rate has surpassed $30 billion, up from about $9 billion at the end of 2025.

The number of enterprise customers spending more than $1 million annually has also doubled to over 1,000 in less than two months, it said.

Most of the new compute infrastructure will be located in the United States, expanding on the company’s previously announced $50 billion investment in domestic computing capacity.

The deal builds on Anthropic’s existing partnerships with Google Cloud and Broadcom, while the company continues to use a mix of hardware platforms, including Amazon’s Trainium chips, Google TPUs and Nvidia GPUs, to train and deploy its models.

Amazon Web Services remains Anthropic’s primary cloud provider and training partner, the company said, adding it is continuing work with AWS on Project Rainier.

Anthropic also said that Claude remains the only frontier AI model available to customers on all three of the world's largest cloud platforms: Amazon Web Services (Bedrock), Google Cloud (Vertex AI), and Microsoft Azure (Foundry).