Amazon has announced plans to invest up to $25 billion more in Anthropic, an artificial intelligence company, in a major expansion of their strategic partnership that also includes a long-term cloud computing commitment expected to exceed $100 billion over the next decade.
The move strengthens Amazon’s position in the fast-growing artificial intelligence and advanced computing market, particularly as demand for large-scale model training and cloud infrastructure continues to accelerate globally.
Under the new arrangement, Amazon will inject $5 billion immediately into Anthropic, with an additional $20 billion contingent on the achievement of agreed commercial milestones. This latest investment adds to the $8 billion Amazon has already committed to the AI startup.
Anthropic, in turn, has pledged to spend more than $100 billion on Amazon Web Services (AWS) over the next 10 years, covering current and next-generation custom chips, including Trainium and Graviton technologies, as well as large-scale compute capacity for training and deploying its AI systems.
According to Amazon, the expanded collaboration builds on a partnership launched in 2023, which has already seen over 100,000 customers using Anthropic’s Claude models through AWS services. Claude has become one of the most widely adopted model families on Amazon Bedrock, the company’s AI inference platform.
The companies are also jointly developing large-scale infrastructure projects, including Project Rainier, described as one of the world’s largest artificial intelligence compute clusters. The system reportedly integrates nearly half a million Trainium2 chips and is being used to train and deploy next-generation Claude models.
Amazon said Anthropic’s long-term commitment to AWS will help scale its training operations while leveraging Amazon’s custom silicon to improve cost efficiency and performance. The agreement also includes expanded international inference capacity across Asia and Europe to support growing global demand for Claude-powered applications.
Andy Jassy, chief executive officer of Amazon, said the collaboration reflects strong demand for its custom AI chips and infrastructure capabilities.
“Our custom AI silicon offers high performance at significantly lower cost for customers, which is why it’s in such hot demand,” Jassy said. He added that Anthropic’s decision to rely on AWS Trainium for the next decade reflects progress in building scalable infrastructure for advanced AI development.
Dario Amodei, chief executive officer and co-founder of Anthropic, said the partnership is essential to meeting rising demand for its models.
“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,” Amodei said, noting that over 100,000 customers are already building on AWS using Anthropic systems.
Beyond funding and infrastructure, the partnership also extends to product integration. AWS customers will be able to access Anthropic’s Claude platform directly through their AWS accounts without requiring separate credentials or billing arrangements. The integration is expected to simplify deployment for developers and enterprises already operating within the AWS ecosystem.
The companies highlighted ongoing collaboration in custom chip development through Annapurna Labs, where Anthropic provides feedback from real-world AI workloads to help shape future generations of Trainium processors. Both firms said this close engineering partnership has accelerated optimisation across training and inference systems.
