OpenAI to sell AI to US agencies through Amazon cloud unit - Yahoo Finance
OpenAI has signed a deal with Amazon Web Services to sell its AI models to U.S. government agencies for classified and unclassified work, the companies confirmed, positioning OpenAI to compete directly for federal contracts that once belonged to Anthropic.
The agreement, first reported by The Information and confirmed to TechCrunch by AWS, allows OpenAI to distribute its models through Amazon Bedrock in government cloud environments including AWS GovCloud and AWS Classified Regions for Secret and Top Secret workloads. The deal builds on an earlier OpenAI contract with the Pentagon announced in late February.
Anthropic had been the primary AI provider for classified defense work under a contract worth up to $200 million, awarded in July 2025. The company used that engagement to set firm boundaries: its contracts with the Pentagon prohibited using its technology for mass domestic surveillance of Americans or fully autonomous weapons. When Anthropic held to those lines, the relationship collapsed. On February 27, President Trump ordered federal agencies to stop using Anthropic products, and Defense Secretary Pete Hegseth formally designated Anthropic a supply-chain risk, blocking contractors from doing business with the company. Anthropic has sued the Pentagon in response.
Into that gap stepped OpenAI. The company had historically focused on unclassified government work. To pursue classified Pentagon contracts, OpenAI restructured its agreement with Microsoft last fall to allow partnerships with rival cloud providers for national security customers, including the Pentagon.
A key detail that separates this from a simple reseller arrangement: OpenAI retains meaningful control. The company decides which models are made available through AWS, receives notice before AWS enables sensitive government customers including intelligence agencies, and can require additional safeguards for specific deployments, according to an OpenAI spokesperson.
AWS, which has invested at least $4 billion in Anthropic, is distributing OpenAI products across its public-sector customer base. Anthropic uses AWS as its primary cloud provider, and Claude models are deeply integrated into Amazon Bedrock for government customers.
The deal also carries commercial logic beyond defense. Companies frequently cite government contracts as a signal of trust and reliability, which could help OpenAI attract enterprise clients watching its federal track record.
What Anthropic refused to compromise on is now OpenAI's entry point. Whether that is pragmatism or something else is a question worth sitting with.