Broadcom Agrees to Expanded Chip Deals with Google and Anthropic

Broadcom announced Monday that it has agreed to produce future generations of artificial intelligence chips for Google and signed an expanded partnership with Anthropic that will give the AI startup access to approximately 3.5 gigawatts of computing capacity drawn from Google's custom AI processors. Broadcom shares climbed more than 3% in extended trading following the disclosure.

The announcement underscores the extraordinary pace at which demand for AI computing infrastructure is accelerating. Anthropic's rise has been particularly striking: the company's annualized revenue has surpassed $30 billion, up from approximately $9 billion at the end of last year, and it now counts more than 1,000 business clients spending over $1 million annually — double the number from just two months ago. The Claude app became the top free app in the U.S. App Store in February, a moment that drove a visible surge in enterprise interest and data center provisioning requests.

The deal deepens an already substantial relationship between Broadcom and Anthropic. On a recent earnings call, Broadcom CEO Hock Tan disclosed that the company was off to a strong start with Anthropic in 2026, providing 1 gigawatt of compute from Google's homegrown tensor processing units (TPUs). Looking to 2027, Tan noted that demand from Anthropic is expected to surge well beyond 3 gigawatts of compute — a figure that analysts at Mizuho translated into roughly $21 billion in AI revenue for Broadcom from Anthropic in 2026, rising to an estimated $42 billion in 2027. The securities filing on Monday did not disclose a specific dollar value for the expanded arrangement.

The broader competitive dynamics in AI compute are rapidly shifting. While model builders like Anthropic and OpenAI continue to rely heavily on Nvidia GPUs through cloud providers such as Amazon, Google, and Microsoft, the move toward custom silicon — including Google's TPUs and chips from Broadcom and AMD — is accelerating as AI companies seek both cost efficiency and differentiated performance at scale. OpenAI has separately committed to drawing on six gigawatts of AMD GPUs, with the first gigawatt expected in the second half of this year. Broadcom is also collaborating with OpenAI on its own custom AI chip, adding further depth to the company's position at the heart of the AI infrastructure arms race.

For investors, the Broadcom-Anthropic-Google deal represents a tangible data point in what has become the defining capital expenditure story of the decade. The infrastructure buildout supporting generative AI is not slowing — it is compounding. And Broadcom, long known as a steady cash-generating semiconductor conglomerate, is increasingly being repriced as a central beneficiary of that trajectory, with AI now commanding the majority of investor attention in the company's quarterly discussions.