Anthropic has signed a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, a supply deal the artificial intelligence company said will help power its frontier Claude models starting in 2027. The expansion is meant to support customer demand worldwide and marks what Anthropic called its most significant compute commitment to date.
Krishna Rao said the partnership continues Anthropic’s disciplined approach to scaling infrastructure and that the company is building the capacity needed to serve exponential growth in its customer base. He said Anthropic is also enabling Claude to define the frontier of AI development, and that the new compute commitment is aimed at keeping pace with unprecedented growth.
The size of that growth is already showing up in Anthropic’s numbers. The company said demand from Claude customers accelerated in 2026, while its run-rate revenue climbed past $30 billion. At the end of 2025, Anthropic said that figure was about $9 billion. It also said the number of business customers spending over $1 million annually has more than doubled in less than two months, rising from over 500 at the time of its Series G fundraising announcement in February to more than 1,000 now.
Read Also: Hock Tan Projects Avgo Stock $100 Billion AI Chip Revenue by 2027
The new agreement is the latest sign that Anthropic’s infrastructure push is moving well beyond a single cloud relationship. The company said the vast majority of the new compute will be sited in the United States, and it described the deal as a major expansion of its November 2025 commitment to invest $50 billion in strengthening American computing infrastructure. It also said the arrangement deepens its existing work with Google Cloud and Broadcom and builds on increased TPU capacity announced last October.
There is a practical wrinkle behind the ambitious language. Anthropic said it trains and runs Claude on AWS Trainium, Google TPUs and NVIDIA GPUs, and that Amazon remains its primary cloud provider and training partner. It also said it continues to work closely with AWS on Project Rainier. The company said Claude remains the only frontier AI model available to customers on Amazon Web Services, Google Cloud and Microsoft Azure, even as it adds another large layer of compute behind the scenes.
Read Also: Anthropic Confirms Is Claude Down With Around 3,000 Downdetector Reports
Anthropic also used the announcement to widen its footprint in the AI ecosystem, saying it is launching the Claude Partner Network and The Anthropic Institute. For now, the clearest takeaway is that the company is betting its next phase of growth will be defined less by software alone than by who can secure the most reliable access to power, chips and cloud capacity.
