Anthropic has entered into a new agreement with Google and Broadcom to secure multiple gigawatts of next-generation TPU capacity, expected to be operational by 2027. This expansion is designed to support the company's Claude models and address the increasing demand from its global customer base.
Strategic Growth: Krishna Rao, CFO of Anthropic, emphasized that this partnership is a continuation of their strategic approach to scaling infrastructure. The new compute capacity is crucial for accommodating the rapid growth in their customer base and advancing AI development.
In 2026, demand from Claude customers surged, leading to a run-rate revenue exceeding $30 billion, a significant increase from approximately $9 billion at the end of 2025. The number of business customers spending over $1 million annually has doubled to over 1,000 in just two months.
Investment in U.S. Infrastructure: The majority of the new compute resources will be located in the United States, furthering Anthropic's commitment to invest $50 billion in American computing infrastructure, initially announced in November 2025.
This partnership enhances Anthropic's existing collaboration with Google Cloud and Broadcom, building on the TPU capacity increase announced last October.
Diverse Hardware Utilization: Anthropic utilizes a variety of AI hardware, including AWS Trainium, Google TPUs, and NVIDIA GPUs, to optimize workload management. This diversity ensures improved performance and resilience for customers relying on Claude for critical tasks. Amazon remains the primary cloud provider and training partner, with ongoing collaboration on Project Rainier.
New Initiatives: Anthropic is also launching the Claude Partner Network to assist enterprises in adopting Claude and establishing The Anthropic Institute to address significant challenges posed by powerful AI technologies.
