Broadcom will produce future generations of artificial intelligence chips for Google, the company said, and will provide Anthropic with computing capacity from Google’s processors.
Broadcom will develop custom tensor processing units for Google, according to a securities filing. Anthropic will access 3.5 gigawatts of compute through Broadcom.
“We have signed a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity that we expect to come online starting in 2027. This significant expansion of our compute infrastructure will power our frontier Claude models and help us serve extraordinary demand from customers worldwide,” wrote Anthropic.
“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,” said Anthropic CFO Krishna Rao.
Under the deal, Broadcom will also supply networking components for Google’s AI server racks until 2031.
The companies did not disclose the financial terms of the agreement. Anthropic’s access to this computing capacity “is dependent on Anthropic’s continued commercial success”, Broadcom wrote in the filing.
Most of the new compute will be located in the United States, Anthropic said. The company trains its Claude AI models on chips from Nvidia and Amazon Web Services, as well as Google’s units.
Anthropic also said today that its revenue run rate had exceeded $30 billion, rising from $9 billion at the end of 2025. It is preparing for an initial public offering, and it has reportedly forecast that its spending on training new AI models will continue to surge through at least 2029.
Broadcom’s (NASDAQ: AVGO) share price closed flat at $314.43, but rose 2.4% after-hours after the deal was announced. Its market capitalisation is $1.49 trillion.
