OpenAI has officially partnered with Amazon Web Services (AWS) to supply its artificial intelligence models to the United States government. This new agreement allows federal agencies and the military to utilize OpenAI’s technology for both classified and unclassified operations.
By leveraging AWS’s extensive cloud infrastructure, the OpenAI AWS government deal positions the ChatGPT creator to expand its footprint in the public sector significantly. The move comes as a major shift in the defense tech landscape, capitalizing on the recent fallout between the Department of Defense and rival AI firm Anthropic.
Expanding the Federal Footprint Through AWS Infrastructure
AWS has confirmed the partnership, which was initially reported by The Information. Through this collaboration, AWS has agreed to distribute OpenAI products across its wide public-sector customer base. Because AWS is already deeply embedded in federal systems, it serves as a natural channel to reach defense and intelligence customers at a massive scale.
The integration includes bringing OpenAI’s models to government cloud environments like AWS GovCloud and AWS Classified Regions. This infrastructure is specifically designed to handle Secret and Top Secret workloads. Previously, OpenAI had limited its government work strictly to unclassified use cases. However, a recent contract with the Pentagon extended its reach into highly classified operations, and the new AWS partnership provides the infrastructure backbone to make that possible.
The Anthropic Controversy and the Pentagon’s Supply Chain Risk
OpenAI’s aggressive push into the government sector fills a sudden gap left by its primary competitor, Anthropic. Amazon had previously invested at least $4 billion in Anthropic, and Anthropic uses AWS as its main cloud provider. Anthropic’s Claude models were deeply integrated into Amazon Bedrock and AWS GovCloud, and the company had worked alongside Palantir and AWS to deploy models in classified military and intelligence systems.
In July 2025, Anthropic successfully secured a Pentagon contract worth up to $200 million. However, that relationship developed severe cracks in February when Anthropic refused to allow unrestricted military use of its artificial intelligence. Specifically, the company pushed back against use cases involving the mass surveillance of American citizens and the development of fully autonomous weapons.
In response to this refusal, the Pentagon officially labeled Anthropic a supply-chain risk, effectively shutting the company out of government work entirely. Following this decision, Anthropic sued the Pentagon. The fallout created an opening in the defense sector, and OpenAI moved quickly to step onto Anthropic’s home turf.
OpenAI’s Safeguards and Deployment Control
Despite the integration with Amazon’s cloud architecture, OpenAI will maintain strict oversight of its technology. The company will retain control over its intellectual property by deciding exactly which models are made available for government use.
Under the terms of the agreement, AWS is required to provide advance notice before enabling access for particularly sensitive government agencies, including intelligence customers. OpenAI plans to coordinate directly with these federal customers to establish deployment terms, security requirements, and operating conditions. The AI firm also holds the right to require additional safeguards for specific deployments.
A Broader $50 Billion Strategic Partnership
This federal distribution agreement follows a monumental partnership announced last month between Amazon and OpenAI. In that broader deal, Amazon committed to a massive $50 billion investment in the AI research organization. The two tech giants revealed plans to co-create a Stateful Runtime Environment powered by OpenAI models. This environment will be available on Amazon Bedrock to help AWS customers build generative AI applications and agents at production scale.
Additionally, the partnership established AWS as the exclusive third-party cloud distribution provider for OpenAI Frontier, a platform that enables organizations to build, deploy, and manage teams of AI agents. To support these advanced workloads, OpenAI is set to consume 2 gigawatts of Trainium capacity through the AWS infrastructure. The companies will also work together to develop customized models that will power Amazon’s own customer-facing applications.
Beyond the immediate public sector applications, securing highly regulated government contracts is often viewed by the broader market as a vital stamp of trust and reliability. Successfully navigating the rigid compliance standards of the US military and intelligence agencies could unlock even more lucrative enterprise contracts for OpenAI in the private sector.
