Anthropic inks deal with Google, Broadcom for TPU capacity

Anthropic said it has inked an agreement with Google and Broadcom to for TPU-powered compute that'll come online in 2027. Anthropic has been expanding its compute infrastructure due to surging demand.

The demand has been so strong that it has had to put caps on Claude Code tokens. That demand is also turning up in Anthropic's annual revenue run rate, which is now $30 billion, up from $9 billion at the end of 2025. Krishna Rao, CFO of Anthropic, said in a blog post that the extra capacity is needed "to serve the exponential growth we have seen in our customer base."

In addition, Anthropic said it has doubled its base of business customers to top 1,000, double the tally from just two months ago. Anthropic added that most of the new compute will be in the US.

Anthropic has a diverse base of suppliers. It works with Google Cloud and Broadcom for TPUs and runs and trains its Claude models on AWS Trainium, Google TPUs and Nvidia GPUs. The company said:

"We train and run Claude on a range of AI hardware—AWS Trainium, Google TPUs, and NVIDIA GPUs—which means we can match workloads to the chips best suited for them. This diversity of platforms translates to better performance and greater resilience for customers who depend on Claude for critical work. Amazon remains our primary cloud provider and training partner, and we continue to work closely with AWS on Project Rainier. 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)."

Google and Broadcom confirmed the Anthropic deal.