Amazon said the Claude artificial intelligence models developed by Anthropic will remain available on its AWS platform for civilian use, except for projects tied to the US Department of Defense. The move follows the Pentagon's designation of Anthropic as a "supply chain risk".

AWS said customers can continue using Claude for any workloads not associated with the Department of Defense, while defense-related projects will have to be moved to other solutions available on the platform. Amazon thus joins Microsoft and Google, which have also confirmed continued access to Anthropic technologies for non-military uses.

Amazon is one of the start-up's leading investors, with $8bn committed since 2023. Anthropic also relies on AWS infrastructure to train its models and plans to use 500,000 Trainium 2 chips as part of its "Project Rainier" data-center initiative, valued at $11bn. Despite the Pentagon's decision, the Claude models remain available on the Bedrock platform for businesses and civilian organizations.