Anthropic has announced a major expansion of its partnership with Google Cloud, revealing plans to scale up its use of Google's advanced Tensor Processing Units TPUs to as many as one million units. The multibillion-dollar initiative, valued at tens of billions of dollars, is expected to bring over one gigawatt of computing capacity online by 2026. This expansion marks one of the largest investments in AI infrastructure to date and will significantly boost Anthropic's ability to conduct cutting-edge research, enhance its AI models, and meet growing customer demand.
The decision builds on a longstanding relationship between Anthropic and Google Cloud, reflecting the strong performance and cost efficiency the company has achieved using Google's AI accelerators. Thomas Kurian, CEO of Google Cloud, emphasized the importance of this collaboration, noting that Anthropic's decision "reflects the strong price-performance and efficiency its teams have seen with TPUs for several years." He added that Google continues to innovate its AI hardware portfolio, including the development of its seventh-generation TPU, Ironwood, to meet the evolving needs of large-scale AI developers.
Anthropic's expansion comes amid rapid business growth, with the company now serving more than 300,000 business customers. Its number of large enterprise clients-each contributing over 100,000 in annual recurring revenue-has surged nearly sevenfold in the past year. The additional compute capacity from Google Cloud will enable Anthropic to support this demand while maintaining the high standards of model quality and reliability that underpin its AI assistant, Claude.
