Anthropic increases computing power deal with Google and Broadcom

AI research lab Anthropic announced on Monday that it has signed a new agreement with Google and Broadcom to increase processing power needed for its Claude AI models. This restructuring of its computing agreements comes as demand for the company's AI models continues to rise.

The agreements will expand Anthropic's use of Google Cloud's tensor processing units, or TPUs, which are the company's advanced AI chips. This expansion is a continuation of a deal struck in October 2025 for over one gigawatt of computing capacity. The new computing capacity is expected to come online in 2027, as stated by Anthropic in a blog post.

While the company did not provide specifics about the computing expansion, a recent SEC filing by Broadcom indicates that the deal includes 3.5 gigawatts of computing. The majority of this computing power will be housed in the U.S. and will be an extension of the company's $50 billion commitment to invest in U.S. computing infrastructure.

Krishna Rao, CFO of Anthropic, stated in the press release, “This groundbreaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure: we are building the capacity necessary to serve the exponential growth we have seen in our customer base while also enabling Claude to define the frontier of AI development.” He also mentioned that the company is making its most significant computing commitment to date to keep pace with its unprecedented growth.

Anthropic did not respond to TechCrunch’s request for comment. The company has seen demand for its Claude models explode in recent months, driven by enterprise customers, despite the U.S. Defense Department labeling Anthropic as a supply chain risk. Additionally, Anthropic recently closed a $30 billion Series G funding round that valued the company at $380 billion.