Broadcom Corporation announced Monday it has entered into an extended partnership with Google to create and manufacture specialized artificial intelligence processors and related hardware for Google’s advanced AI systems until 2031.
Additionally, the technology firm reached a separate agreement with AI company Anthropic, granting the startup access to approximately 3.5 gigawatts of artificial intelligence computing power using Google’s processors, beginning in 2027.
Neither company revealed the monetary value of these partnerships.
Following the announcement, Broadcom’s stock price climbed roughly 3% during after-hours trading.
The market for specialized processors like Google’s tensor processing units (TPUs), designed specifically for artificial intelligence tasks, has grown dramatically as companies look for cost-effective alternatives to Nvidia’s expensive graphics processing units.
Previous reports from December indicated Google has been working to position its TPUs as competitive options against Nvidia’s dominant graphics processors. Sales of these tensor processing units have emerged as a vital component of Google’s cloud computing revenue growth, helping demonstrate to shareholders that artificial intelligence investments are producing financial returns.
Anthropic stated Monday that this latest agreement supports the company’s pledge to invest $50 billion in bolstering American computing infrastructure.
The AI startup reported that interest in its Claude artificial intelligence model has grown rapidly in 2026, with annual revenue projections now exceeding $30 billion, compared to approximately $9 billion recorded at the close of 2025.
According to Anthropic, the company develops and operates Claude using various AI hardware platforms, including Amazon Web Services’ Trainium processors, Google’s tensor processing units, and Nvidia’s graphics processors.
Amazon continues to serve as Anthropic’s main cloud computing provider and development partner.
