Anthropic has agreed to access about 3.5 gigawatts of Google’s next-generation TPU processor computing power through Broadcom starting in 2027, its largest infrastructure commitment to date — while also revealing that its revenue has surpassed $30 billion, tripling from about $9 billion at the end of 2025.
See also: Anthropic acquired a startup with fewer than 10 people
Anthropic announced that it is securing multiple gigawatts of next-generation computing power through a new agreement with Google and Broadcom, while revealing revenue growth figures that highlight why the AI lab now requires infrastructure on a scale that would have seemed impossible just two years ago. The agreement, announced on April 6, 2026, gives Anthropic access to approximately 3.5 gigawatts of Google TPU power through Broadcom starting in 2027, expanding on the 1 gigawatt already provided to the company in 2026.
Krishna Rao, Anthropic’s chief financial officer, described it as “the company’s most significant computing power commitment to date,” adding that the deal represents a continuation ofthe company’s “disciplined approach to scaling infrastructure.” The majority of the new capacity will be located in the United States, expanding on Anthropic’s November 2025 commitment to invest $50 billion in U.S. AI computing power.
The announcement is as much about Broadcom as it is about Anthropic or Google. Under the new deal, Broadcom will act as the middleman between Google’s custom silicon and Anthropic’s training and prediction workloads. In addition, Broadcom has signed a separate long-term agreement with Google to design and supply future generations of custom TPU chips, as well as a supply assurance agreement to provide networking and other components for Google’s next-generation AI data racks through 2031.
See also: Claude Code discovered zero-day in Vim and GNU Emacs
That makes Broadcom an increasingly necessary node in the AI infrastructure graph. The chipmaker, led by CEO Hock Tan, doesn’t build AI models. It builds the silicon and interconnects on which AI models are built. Broadcom shares rose about 3% in extended trading after the announcement, a reaction that reflects investor appetite for companies that are at the physical layer of the AI stack rather than the application layer above it.
Mizuho analysts , led by Vijay Rakesh , estimated that Broadcom will record $21 billion in AI revenue from Anthropic in 2026 alone, rising to $42 billion in 2027 , numbers that, even as projections, illustrate the financial weight of what is being committed.
When the Series G closed, Anthropic reported that more than 500 enterprise customers were spending more than $1 million annually. Since the April announcement, that number has surpassed 1,000, doubling in less than two months. The pace of enterprise adoption is a direct driver of the expansion of computing power: more revenue requires more predictive power, more predictive power requires more training compute, and more training compute requires more gigawatts.
What distinguishes Anthropic’s infrastructure approach from many of its peers is an explicit multi-vendor chip strategy. Claude is trained and served on three hardware platforms: Amazon’s Trainium chips, Google’s TPUs , and Nvidia’s GPUs . Anthropic says that Claude is the only cutting-edge model available on all three major cloud platforms, AWS, Google Cloud, and Microsoft Azure, a claim that carries commercial as well as technical significance.
See also: Google DeepMind: How do hackers breach AI Agents?
The multi-vendor stance gives Anthropic both resilience and bargaining power. If capacity is limited on any one platform, workloads can be moved. If a chip manufacturer faces supply disruption, export controls, or pricing pressures, Anthropic is not fully exposed to that shock.
