Anthropic just locked in its biggest infrastructure deal yet: multiple gigawatts of next-generation TPU capacity from Google and Broadcom, starting in 2027. CFO Krishna Rao called it their "most significant compute commitment to date." The numbers explain why. Anthropic's run-rate revenue hit $30 billion, up from $9 billion just months ago. They've doubled their big-spending enterprise customers in under two months. Over 1,000 businesses now spend more than $1 million annually on Claude. A single gigawatt powers roughly 750,000 homes, so Anthropic is essentially building the electrical footprint of a small city to train and run Claude. That kind of growth demands infrastructure at a scale few companies can even contemplate. These capacity constraints have already forced the company to block external agents.
Amazon remains Anthropic's primary cloud partner. The company still trains on AWS Trainium chips and NVIDIA GPUs alongside Google's TPUs, matching workloads to whatever hardware handles them best. Claude is the only frontier model available on all three major clouds: AWS Bedrock, Google Vertex AI, and Microsoft Azure. Most of this new capacity will sit in the US, part of a $50 billion commitment to American infrastructure Anthropic made in late 2025.