Amazon is once again making a substantial investment in Anthropic: $5 billion upfront, with the potential for an additional $20 billion. At the same time, Anthropic has committed to spending more than $100 billion on AWS over the next 10 years, including up to 5 gigawatts of Trainium capacity for training and running Claude models.
This involves Trainium2, Trainium3, and Trainium4 chips, as well as tens of millions of Graviton cores. Notably, Anthropic aims to secure up to 5 gigawatts (GW) of computing capacity for training and running Claude models.
The foundation for this expansion is Project Rainier. With nearly half a million Trainium2 chips, it is one of the largest AI computer clusters in the world. Anthropic is already actively using it to train Claude models. Significant Trainium3 capacity is planned for later this year.
More than 100,000 organizations are already running Claude via Amazon Bedrock. Dario Amodei, CEO and co-founder of Anthropic, states: “Our users tell us Claude is increasingly essential to how they work, and we need to build the infrastructure to keep pace with rapidly growing demand.”
Claude Platform Now Directly Available via AWS
AWS customers can access Anthropic’s native Claude experience from their existing AWS account, without additional login credentials or separate billing. The partnership dates back to 2023, when Amazon first invested $4 billion, and AWS became Anthropic’s primary cloud provider. The rollout of Claude has since expanded to Asia and Europe.