AI BriefingAnthropic, Massive Compute Expansion
Key point
Anthropic unveiled a massive compute expansion spanning Google TPUs, Azure, and Broadcom.
Details
Anthropic has announced a series of large-scale compute expansions centered on Google Cloud TPUs, Fluidstack data centers, Microsoft Azure, NVIDIA, and Broadcom.
- October 23, 2025: Announced plans to expand Google Cloud TPUs to up to 1 million units and bring over 1GW of capacity online in 2026.
- November 12, 2025: Announced a $50B investment in US AI infrastructure across Texas and New York, expected to create 800 permanent jobs and 2,400 construction jobs.
- November 18, 2025: Unveiled a strategic partnership with Microsoft and NVIDIA involving a $30B Azure compute purchase and capacity agreements for up to an additional 1GW, expanding Claude to Microsoft Foundry and the Copilot family.
- April 6, 2026: Signed a next-generation TPU agreement of several GW in scale with Google and Broadcom, to be deployed starting in 2027, with Broadcom's 8-K referencing approximately 3.5GW.
Anthropic continues to maintain a multi-chip strategy using TPUs, AWS Trainium, and NVIDIA GPUs together, keeping AWS as its primary training partner while scaling infrastructure to meet growing demand for Claude.
Claude is available across AWS Bedrock, Google Cloud Vertex AI, and Microsoft Azure Foundry.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Google Launches First Suncatcher Orbital AI Data Center Test on October 1Finnish AI Cloud Startup Verda Raises $189 Million in Series B