Anthropic has reached an agreement to expand its use of Google Cloud’s tensor processing units (TPUs), securing access to substantial computing capacity that the company will use to train and serve future Claude models.

The capacity will include more than one Gigawatt (1GW) coming online in 2026 and access to up to one million TPU chips, together with additional Google Cloud services.

CNBC reported the deal is worth tens of billions of dollars and that this is Anthropic’s largest TPU commitment to date.

Anthropic said the expanded capacity will supply the compute resources required for its research and development (R&D) teams to train and deploy subsequent Claude models.

Google Cloud CEO Thomas Kurian said: “Anthropic’s choice to significantly expand its usage of TPUs reflects the strong price-performance and efficiency its teams have seen with TPUs for several years.

“We are continuing to innovate and drive further efficiencies and increased capacity of our TPUs, building on our already mature AI accelerator portfolio, including our seventh generation TPU, Ironwood.”

The agreement extends a strategic partnership first announced in 2023. At that time, Anthropic began using Google Cloud’s AI infrastructure to train models and made them available through Google Cloud’s Vertex AI platform and the Google Cloud Marketplace.

Thousands of businesses are said to use Claude models on Google Cloud, including Figma, Palo Alto Networks, and Cursor.

Anthropic CFO Krishna Rao said: “Anthropic and Google have a longstanding partnership and this latest expansion will help us continue to grow the compute we need to define the frontier of AI.

“Our customers—from Fortune 500 companies to AI-native startups—depend on Claude for their most important work, and this expanded capacity ensures we can meet our exponentially growing demand while keeping our models at the cutting edge of the industry.”

Recently, Google Cloud launched G4 virtual machines (VMs) powered by Nvidia RTX PRO 6000 Blackwell Server GPUs.