Amazon has agreed to invest up to $25 billion in Anthropic, the San Francisco-based AI startup behind the Claude family of models.
The deal also comes with a commitment from Anthropic to spend more than $100 billion on Amazon Web Services technologies over the next decade.
The investment structure splits into two tranches: $5 billion goes to Anthropic immediately, with the remaining $20 billion contingent on meeting specific commercial milestones.
The deal was struck at a valuation of $350 billion, not including the new funding, a notably lower figure than the $380 billion valuation Anthropic commanded in a February funding round.
Amazon’s history with Anthropic stretches back to September 2023, when it made an initial $4 billion investment and became the startup’s primary cloud partner.
A second $4 billion tranche followed in November 2024, making AWS Anthropic’s primary training partner as well. Monday’s announcement brings Amazon’s total commitment to $33 billion.
A chip deal at the heart of the partnership
Beyond the funding, the agreement is as much about hardware as it is about capital.
Anthropic has committed to running its large language models on AWS Trainium Amazon’s custom AI chips for the next ten years, spanning current and future chip generations.
The deal secures Anthropic up to 5 gigawatts of capacity on Amazon’s Trainium chips.
Nearly 1 gigawatt of Trainium2 and Trainium3 capacity is expected to come online by the end of this year.
Anthropic currently uses more than 1 million Trainium2 chips to train and serve Claude.
Amazon CEO Andy Jassy framed the chip commitment as a validation of the company’s silicon strategy.
“Our custom AI silicon offers high performance at significantly lower cost for customers, which is why it’s in such hot demand,” Jassy said in a statement.
The compute expansion appears tied to a concrete business problem. Anthropic noted that recent surges in consumer and enterprise demand for Claude had placed significant pressure on existing systems, leading to reliability issues that the new Amazon agreement aims to resolve.
Anthropic CEO Dario Amodei added that Claude had become “increasingly essential” to how users work, and that the company needed to build infrastructure to keep pace with rapidly growing demand.
Amazon plays the same hand twice
The move mirrors a nearly identical arrangement Amazon struck two months earlier with OpenAI: up to $50 billion in investment and a $100 billion cloud commitment.
The parallel deals position AWS as the infrastructure backbone for the world’s two leading AI labs, a calculated bet that whoever wins the AI race will need Amazon’s cloud underneath them.
The strategy matters because Amazon has struggled to generate significant buzz around its own AI models, such as Nova, while continuing to lead in providing critical infrastructure for the AI boom.
Backing both Anthropic and OpenAI lets Amazon hedge on the model layer while cementing AWS as the default compute provider.
The Anthropic deal also carries a competitive sting for OpenAI.
OpenAI executives had recently criticized Anthropic for making a “strategic misstep to not acquire enough compute” and for operating on a meaningfully smaller curve.
The 5-gigawatt capacity guarantee is, in part, a direct answer to that claim.
What it means for AWS customers
For businesses already using AWS, the expanded partnership means deeper access to Anthropic’s models through
Amazon Bedrock, Amazon’s managed service for building generative AI applications.
Enterprises can deploy Claude directly within their existing cloud infrastructure. That removes a significant integration barrier and keeps AI workloads within a single platform.
Amazon shares rose around 2.7% in extended trading following the announcement.
The deal reflects a broader dynamic reshaping the technology industry: AI development is consolidating around a small number of cloud providers and well-funded startups, with hundreds of billions of dollars being deployed to secure long-term positions before the competitive landscape fully settles.
