Anthropic expands Amazon tie-up in USD $5 billion deal
Anthropic has expanded its collaboration with Amazon in a deal for up to 5 gigawatts of new computing capacity. The agreement also includes a new USD $5 billion Amazon investment in the artificial intelligence company.
Under the arrangement, Anthropic will commit more than USD $100 billion over the next 10 years to Amazon Web Services technologies to secure capacity for training and running its Claude models. The commitment covers Graviton processors and Trainium chips from Trainium2 through Trainium4, with an option to buy future generations of Amazon custom silicon.
New Trainium2 capacity is due to come online in the first half of the year, with nearly 1 gigawatt of total Trainium2 and Trainium3 capacity expected by the end of 2026. The added infrastructure will also support Claude on Amazon Bedrock, including expanded inference capacity in Asia and Europe.
Deeper ties
The latest agreement extends a relationship that began in 2023. More than 100,000 customers already use Claude on Amazon Bedrock, and the companies have also launched Project Rainier, described as one of the largest compute clusters in the world.
Anthropic currently uses more than one million Trainium2 chips to train and serve Claude. AWS will remain its primary training and cloud provider for mission-critical workloads.
Amazon's new USD $5 billion investment adds to the USD $8 billion it has already put into Anthropic. It may invest up to a further USD $20 billion in the future.
The deal also brings the full Claude Platform directly into AWS. Customers will be able to use the platform within AWS accounts with the same controls and billing arrangements, without additional credentials or contracts.
That move broadens Anthropic's distribution at a time when access to large-scale computing infrastructure has become a major constraint for AI model developers. Demand from enterprise customers, developers and consumers has risen sharply this year.
Demand surge
Anthropic said its run-rate revenue has surpassed USD $30 billion, up from about USD $9 billion at the end of 2025. It also said strong consumer growth across its free, Pro and Max tiers has affected reliability and performance for free, Pro, Max and Team users, particularly during peak hours.
The Amazon agreement is expected to expand available capacity quickly, with meaningful compute arriving within the next three months. Those near-term additions, together with broader hardware diversification, are intended to ease pressure from rapid usage growth.
For Amazon, the agreement strengthens demand for its in-house AI chips as it competes with other cloud groups for model training and inference workloads. Claude will remain available across AWS, Google Cloud and Microsoft Azure, but AWS will continue as Anthropic's main provider for training and cloud infrastructure.
Andy Jassy, Chief Executive Officer of Amazon, said the deal reflected growing uptake of the company's custom AI silicon. "Our custom AI silicon offers high performance at significantly lower cost for customers, which is why it's in such hot demand," he said. "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, as we continue delivering the technology and infrastructure our customers need to build with generative AI."
Anthropic presented the expanded arrangement as a response to growing reliance on Claude in daily work. Dario Amodei, Chief Executive Officer and Co-Founder of Anthropic, said the company needed more infrastructure to keep up with demand. "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," he said. "Our collaboration with Amazon will allow us to continue advancing AI research while delivering Claude to our customers, including the more than 100,000 building on AWS."