Anthropic signs multi-gigawatt TPU deal with Google and Broadcom

Anthropic announced on April 6, 2026 that it has signed a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity. The company said the new infrastructure is expected to begin coming online in 2027 and will be used to power future Claude models as customer demand continues to accelerate.

The scale matters because Anthropic is explicitly describing this as its biggest compute commitment yet. In the same announcement, the company said its run-rate revenue has surpassed $30 billion, up from approximately $9 billion at the end of 2025. It also said the number of business customers spending more than $1 million on an annualized basis has climbed from more than 500 in February to more than 1,000 now, doubling in less than two months.

Anthropic added that the vast majority of the new compute will be located in the United States. That makes the deal a significant expansion of its November 2025 commitment to invest $50 billion in American computing infrastructure. The company also said the agreement deepens its existing Google Cloud TPU relationship and its ties with Broadcom, even as Amazon remains Anthropic’s primary cloud provider and training partner through Project Rainier.

Strategically, the announcement shows how the frontier model race is increasingly constrained by access to power, datacenter buildout, and specialized chips rather than by model ideas alone. Anthropic emphasized that it continues to run Claude across multiple hardware stacks, including AWS Trainium, Google TPUs, and NVIDIA GPUs. That hardware diversity is meant to improve resilience and allow different workloads to land on the most suitable accelerators.

For the wider AI market, the message is clear: compute supply remains a central competitive moat. A multi-gigawatt TPU agreement is not just an infrastructure procurement story. It is also a signal that leading model labs are locking in long-range capacity years ahead, tying together cloud partners, chip designers, and national industrial policy as they prepare for larger and more expensive AI systems.