Broadcom agrees to expanded chip deals with Google, Anthropic

In a significant development for the tech industry, Broadcom announced on Monday its agreement to manufacture advanced artificial intelligence chips for Google. Additionally, the company has enhanced its partnership with Anthropic, providing the AI startup access to approximately 3.5 gigawatts of computing power utilizing Google's AI processors. Following this announcement, Broadcom's shares experienced a 3% increase during after-hours trading, while Alphabet, Google's parent company, saw its stock remain stable. This collaboration highlights the escalating need for robust infrastructure capable of supporting generative AI models, underscoring the growing demand for such technologies. Anthropic has gained notable traction this year, with its Claude application reaching the top of the free U.S. app rankings on Apple's App Store in February, especially after its public dispute with the Pentagon. In December, Broadcom's CEO Hock Tan revealed that Anthropic had placed a substantial order worth $10 billion for customized chips. However, the recent filing did not specify any financial details regarding the new agreements. During a recent earnings call, Tan mentioned that for 2026, they anticipate starting off strong with one gigawatt of TPU computing capacity for Anthropic, with expectations for demand to exceed three gigawatts by 2027. Concurrently, Broadcom is also working with OpenAI, a competitor of Anthropic, to develop specialized silicon for AI applications. Currently, both Anthropic and OpenAI heavily rely on Nvidia's graphics processing units through major cloud service providers such as Amazon, Google, and Microsoft. Furthermore, OpenAI has pledged to utilize six gigawatts of AMD's GPUs, with the first gigawatt expected to be operational in the latter half of this year.