Amazon Keeps Claude AI Available, Except for Military Projects

Amazon confirms that Anthropic’s artificial intelligence technology, Claude, will remain available to customers of its cloud computing service Amazon Web Services (AWS). However, the usage does not extend to workloads related to the United States Department of Defence (DoD).

An AWS spokesperson said in a formal statement that ‘customers and AWS partners will still be able to use Claude for all workloads not related to the Department of War (DoW).’

The statement came after US President Donald Trump, via a post on social media last week, ordered federal agencies to halt the use of Anthropic’s technology. The order followed Anthropic’s refusal of the DoD’s request to have the right to operate the company’s technology across all legitimate uses without restrictions.

With the supply-chain risk label in place, vendors and defence contractors are required to ensure they do not use Anthropic’s AI model in Pentagon-related work.

Microsoft, on Thursday night (5 March 2026), said that Anthropic’s Claude model remains accessible in its products as long as it is not used for defence work. Google also issued a similar statement on Friday.

Amazon itself is one of Anthropic’s largest investors, having invested around $8 billion in the startup since 2023. The two firms have also forged a close commercial relationship.

AWS remains the startup’s main partner for cloud computing services and AI model training.

Anthropic has also committed to using 500,000 Amazon-made chips, specifically Trainium 2, as part of building AWS’s data centre campus valued at $11 billion for the startup, known as Project Rainier.

AWS also provides Bedrock service for GovCloud, a cloud region designed to store sensitive data and run tightly controlled workflows.