Semiconductor company Broadcom announced on Monday it has agreed to produce future versions of Google’s artificial intelligence chips. In a securities filing, the company also detailed an expanded partnership with AI firm Anthropic to provide it with access to computing capacity that utilizes Google’s processors.
Following the announcement, Broadcom’s shares rose 3% in extended trading.
The new agreement will see Broadcom manufacture future generations of Google’s Tensor Processing Units (TPUs), which are custom-designed chips for AI workloads. The expanded deal with Anthropic will provide the AI startup with access to about 3.5 gigawatts of computing capacity powered by these TPUs. The filing did not disclose the financial terms of the agreements.
The announcement provides further detail to comments made by Broadcom CEO Hock Tan during an earnings call last month. On the call, Tan stated that for its work with Anthropic, Broadcom was “off to a very good start in 2026” in providing 1 gigawatt of compute from Google’s TPUs. He added that for 2027, “this demand is expected to surge in excess of 3 gigawatts of compute.”
In a note published after that earnings call, analysts from Mizuho led by Vijay Rakesh estimated that the partnership with Anthropic could generate $21 billion in AI revenue for Broadcom in 2026 and $42 billion in 2027.
The agreements come as technology companies work to secure the specialized infrastructure required to develop and run large-scale generative AI models. AI developers like Anthropic and its competitor, OpenAI, currently rely on graphics processing units (GPUs) supplied primarily by Nvidia through cloud providers such as Google, Amazon, and Microsoft.
In addition to its work with Google and Anthropic, Broadcom is also collaborating with OpenAI on developing custom silicon for AI. Other chipmakers are also involved in supplying the sector, with OpenAI having committed to using AMD’s GPUs.
