Anthropic Secures Multiple Gigawatts of Next-Generation TPU Capacity from Google and Broadcom

TLDR\n\nAnthropic signed a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU-based AI compute capacity, with the vast majority sited in the United States and expected to come online starting in 2027. This expands Anthropic’s existing use of Google Cloud TPUs and services. The deal supports rapid scaling of frontier Claude models as the company’s run-rate revenue surpasses $30 billion (up from ~$9 billion at end of 2025) and its number of business customers spending over $1 million annualized exceeds 1,000 (doubled since February).\n\n## Partnership Details\n\nThe agreement builds on prior collaborations:\n- Anthropic will access multiple gigawatts of TPU capacity through Google Cloud services and Google-built TPUs supplied via Broadcom.\n- Capacity rollout begins in 2027.\n- Part of Anthropic’s $50 billion commitment to American AI infrastructure announced in November 2025.\n- Most new capacity will be located in the US.\n\nAnthropic continues its multi-cloud strategy, with Amazon Web Services as primary cloud provider and training partner (including Project Rainier), alongside Google Cloud (Vertex AI) and Microsoft Azure (Foundry). Claude remains the only frontier model available across all three major clouds.\n\n## Business Context\n\nAnthropic’s growth has accelerated sharply:\n- Run-rate revenue now exceeds $30 billion, up from approximately $9 billion at the end of 2025.\n- Business customers each spending over $1 million annualized doubled from over 500 (at Series G in February) to more than 1,000 in under two months.\n- Demand spans foundation models, agents, and enterprise applications.\n\nCFO Krishna Rao: “This groundbreaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure: we are building the capacity necessary to serve the exponential growth we have seen in our customer base while also enabling Claude to define the frontier of AI development.”\n\n## Why this story matters\n\nThe deal represents one of the largest single compute commitments by a frontier AI lab, underscoring the massive infrastructure requirements for next-generation models. It highlights the strategic importance of custom silicon (TPUs) and diversified hardware ecosystems for performance and resilience. By siting capacity in the US, it aligns with national priorities around AI infrastructure leadership. The rapid customer and revenue growth signals strong enterprise adoption of frontier AI, while the multi-cloud approach gives customers flexibility across providers.\n\n## Sources\n- Anthropic official announcement: “Anthropic expands partnership with Google and Broadcom for multiple gigawatts of next-generation compute” (April 6, 2026). https://www.anthropic.com/news/google-broadcom-partnership-compute\n- Google Cloud press release: “Anthropic Expands Use of Google Cloud and TPUs” (April 6, 2026). https://www.googlecloudpresscorner.com/2026-04-06-Anthropic-Expands-Use-of-Google-Cloud-and-TPUs\n- Broadcom Inc. Form 8-K filing (April 6, 2026), detailing the Long Term Agreement for TPUs and Anthropic’s ~3.5 GW allocation.\n- Anthropic Series G announcement context (February 2026) for customer and revenue baselines.\n\n## Featured Image Alt Text\n\nAnthropic logo alongside Google TPU chips and Broadcom networking hardware in a data center rack, symbolizing the multi-gigawatt TPU compute partnership announced April 6, 2026\n\n## Tags\nAnthropic, Google, Broadcom, TPU, Compute, Infrastructure, AI Scaling, Multi-Cloud, Claude