Summary
Anthropic said it will spend more than $100 billion over ten years on AWS infrastructure and secure up to 5GW of capacity for training and deploying Claude. The deal includes Trainium2 through Trainium4 roadmap access, nearly 1GW of Trainium2 and Trainium3 capacity by the end of 2026, regional inference expansion, and a coming Claude Platform on AWS.
What changed
Anthropic expanded its AWS relationship into a quantified long-term compute deal covering up to 5GW of capacity, Trainium2 through Trainium4 roadmap access, and broader regional inference expansion.
Why it matters
This is one of the clearest public infrastructure commitments from a frontier-model vendor. It gives enterprises a stronger read on Anthropic's capacity outlook while also underscoring how tightly Claude's scale path is now linked to AWS, Trainium, and a more native AWS distribution path.
Evidence excerpt
Anthropic says over 100,000 customers now run Claude on Bedrock, the new agreement secures up to 5GW of capacity, and Claude Platform on AWS is coming soon.