Anthropic has struck a major infrastructure deal with Google and Broadcom to secure multi-gigawatt computing capacity, as demand for its Claude AI models continues to climb.
Details disclosed in a recent securities filing show the semiconductor firm will support future iterations of Google’s tensor processing units, with part of that capacity allocated to Anthropic. The arrangement is expected to unlock roughly 3.5 gigawatts of compute, with deployments set to begin scaling from 2027.
Anthropic said its annualized revenue has now crossed $30 billion, up sharply from around $9 billion at the end of last year. The company also reported that more than 1,000 enterprise customers are each spending over $1 million annually, a figure that has doubled within weeks.
“We are making our most significant compute commitment to date to keep pace with our unprecedented growth,” Anthropic’s chief financial officer Krishna Rao said, adding that the partnership would “build the capacity necessary to serve the exponential growth we have seen in our customer base.”
Most of the new infrastructure will be based in the United States, extending an earlier pledge to invest $50 billion into domestic compute capacity. The expansion also builds on Anthropic’s existing relationships with Google Cloud and Broadcom, following earlier TPU capacity announcements.
From Broadcom’s side, the deal adds to a growing pipeline of AI-linked revenue. CEO Hock Tan had previously confirmed that the company was already supplying around 1 gigawatt of compute for Anthropic through Google’s TPU systems, with demand expected to climb past 3 gigawatts in 2027.
For Broadcom, the latest deal adds to a quickly growing list of AI infrastructure partnerships. During the company’s March earnings call, Broadcom CEO Hock Tan said it was already supplying roughly 1 gigawatt of compute for Anthropic and added that this was expected to surpass 3 gigawatts by 2027.
Wall Street estimates suggest the partnership could translate into significant earnings. Analysts at Mizuho have projected that Broadcom may generate about $21 billion in AI-related revenue from Anthropic in 2026, potentially doubling to $42 billion the following year.
At the same time, competition across AI infrastructure remains intense. AI developers, including Anthropic and its peers, continue to rely on a mix of hardware platforms, including Nvidia GPUs, Google TPUs, and custom chips.
Broadcom is also working with OpenAI on separate silicon efforts, while cloud providers such as Amazon, Google, and Microsoft remain central to delivering that compute at scale.
Anthropic noted that its Claude models are now deployed across all three major cloud ecosystems, allowing workloads to be distributed depending on performance needs.