Under the expanded agreement, Anthropic has committed to spending more than $100 billion over the next 10 years on AWS technology, including current and future generations of Amazon custom AI chips such as Trainium and Graviton processors.
Amazon and humane announced a significant expansion of its strategic partnership, deepening collaboration across artificial intelligence infrastructure, model development, and cloud services.
The companies said they have been working together since 2023 to accelerate the adoption of generative AI across industries, enabling customers to build and scale AI applications. Over 100,000 customers are currently running Anthropic’s Claude models on AWS, making Claude one of the most widely used model families on Amazon Bedrock. The companies are also collaborating on Project Rainier, a large-scale AI computing cluster designed to support advanced model training and deployment.
Under the expanded agreement, Anthropic has committed to spending more than $100 billion over the next 10 years on AWS technology, including current and future generations of Amazon custom AI chips such as Trainium and Graviton processors. The partnership will also provide up to 5 gigawatts of computing capacity to support training and running advanced AI models, with significant new Trainium3 capacity expected to come online this year. This partnership will further expand international AI inference capabilities across Asia and Europe, meeting the growing global demand for Claude models.
The agreement also covers future chip generations, including Trainium2, Trainium3, and Trainium4, as Anthropic continues to scale its AI workloads using AWS infrastructure. The companies noted that AWS’ Trainium and Graviton technologies already support more than 100,000 customers and power most inference workloads on Amazon Bedrock.
The new integration allows AWS customers to access Anthropic’s Claude platform directly through their AWS accounts using their existing credentials, security controls, and billing systems. Depending on your needs, customers can choose between Claude on Amazon Bedrock or the Claude native experience on AWS.
As part of the expanded relationship, Amazon will invest $5 billion in Anthropic, with an additional $20 billion based on future commercial milestones. This builds on $8 billion in investments to date.
“Our custom AI silicon is in great demand because it delivers high performance at a significantly lower cost for our customers. Anthropic’s commitment to running language models at scale on AWS Trainium over the next decade reflects the progress we’ve made together in custom silicon as we continue to deliver the technology and infrastructure our customers need to build with generative AI.”
– Andy Jassy, Amazon CEO
“Our users tell us that Claude has become increasingly essential to the way they work, and they need to build infrastructure to meet rapidly growing demand. Our collaboration with Amazon allows us to continue advancing our AI research while delivering Claude to our customers, including over 100,000 builds on AWS.”
– Dario Amodei, Anthropic CEO and Co-Founder
The companies highlighted several key joint initiatives, including close collaboration on custom silicon development with Amazon’s Annapurna Labs, where Anthropic will provide direct feedback to shape future AI chip designs. Another major initiative, Project Rainier, is said to be one of the world’s largest AI computing clusters, with nearly half a million Trainium2 chips supporting large-scale training and deployment of cloud models.
Amazon and Anthropic also noted increased adoption of Claude by businesses across industries. Examples include how Lyft used Claude via Amazon Bedrock to automate customer support and reduce resolution time by 87%, and how Pfizer leveraged Claude to streamline research workflows, saving thousands of scientists’ time annually while reducing infrastructure costs.
The companies reaffirmed their commitment to making AWS the primary platform for training and deploying Anthropic’s large-scale models, while also enabling Amazon teams to use cloud models to improve services across the broader ecosystem.
