Amazon has committed to investing $5 billion in Anthropic, with a potential additional investment of up to $20 billion tied to commercial targets. This investment raises Amazon’s total potential investment in Anthropic to $33 billion, following a previous commitment of $8 billion.
The expanded partnership includes an agreement for Anthropic to secure up to 5 gigawatts (GW) of Amazon’s Trainium chips, which are expected to support the development and deployment of Anthropic’s AI models. Anthropic will also invest over $100 billion in AWS technologies over the next decade, covering current and future Trainium chips and millions of Graviton CPU cores.
Anthropic’s AI models, particularly the Claude family, will utilize these chips for training and inference workloads. Significant capacity of the Trainium3 chip is anticipated to be available later this year as part of the agreement. Both companies plan to enhance international inference capabilities across Asia and Europe for Anthropic’s expanding global user base.
More than 100,000 organizations are reported to use Anthropic’s Claude models through AWS, making it a widely adopted offering on Amazon’s Bedrock inference service. Anthropic CEO Dario Amodei stated, “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.”
AWS primarily utilizes Trainium chips for its inference workloads. Both Trainium and Graviton technologies support over 100,000 customers. Anthropic relies on AWS as its main cloud and training provider for essential workloads.
Under the new agreement, AWS customers can access Anthropic’s full Claude Platform directly through their AWS accounts. This integration allows users to leverage AWS’s existing access controls and monitoring systems, eliminating the need for separate contracts or credentials.
Amazon CEO Andy Jassy remarked, “Anthropic’s commitment to run its large language models on AWS Trainium for the next decade reflects the progress we’ve made together on custom silicon.”
The partnership also focuses on large-scale infrastructure projects like Project Rainier, which features one of the world’s largest AI compute clusters with approximately 500,000 Trainium2 chips, used for training and deploying Claude models globally. Anthropic collaborates with Amazon’s Annapurna Labs to provide feedback on Trainium chip design to better address the needs of advanced AI models.
