March 17 (Reuters) - OpenAI has signed a new deal to sell access to its AI models to U.S. defense and government agencies through Amazon's cloud unit for classified and unclassified work, the ChatGPT maker said on Tuesday.
The contract enables OpenAI to support the Pentagon under a deal it secured late last month, after the agency dropped its previous AI provider, Anthropic.
Anthropic, which won a Pentagon contract worth up to $200 million in July 2025, had been a key U.S. defense AI supplier, working
with Palantir and Amazon Web Services (AWS) to deploy its Claude models in classified military and intelligence systems.
But its relationship with the Pentagon collapsed in February after Anthropic refused to allow unrestricted military use of its AI, particularly for domestic surveillance and autonomous weapons. Following which, the Pentagon labeled it a "supply chain risk" and effectively cut it off from government work.
OpenAI, which had previously focused on unclassified government use, has now secured a Pentagon contract to provide its models for classified operations.
Its partnership with AWS builds on this latest shift, reflecting how access to government and defense contracts, especially via cloud providers already embedded in federal systems, is becoming a key battleground.
Securing government contracts could also help the ChatGPT maker attract large corporate clients, which often see high-stakes public sector work as a strong signal of trust and reliability.
Following OpenAI's transition to a for-profit structure last fall, the company updated its agreement with Microsoft to allow partnerships with rival cloud providers in selling AI to national security customers, including the Pentagon.
(Reporting by Akash Sriram in Bengaluru; Editing by Anil D'Silva and Shinjini Ganguli)
