The Pentagon has announced agreements with seven prominent artificial intelligence companies to implement their technologies within classified military systems. This strategic move notably excludes Anthropic, which is currently in conflict with the Defense Department over the regulations surrounding the usage of its AI tools.
The companies included in the agreements are SpaceX, OpenAI, Google, Nvidia, Reflection, Microsoft, and Amazon Web Services. These organizations will be integrated into the Pentagon’s most sensitive classified networks, which are utilized for mission planning, weapons targeting, and various military operations. Meanwhile, Anthropic’s absence from the announcement signals a continued rift, as the company has voiced concerns regarding the use of its technology in mass domestic surveillance and the control of lethal autonomous weapons.
Previously, the US military had begun to reduce its reliance on Anthropic’s AI following assertions from the company that its technologies should not be used in ways that could compromise ethical standards. As a consequence of these disagreements, the Pentagon designated Anthropic as a supply-chain risk earlier this year, effectively barring its application by military entities and contractors. In February, President Trump instructed that the government “immediately cease” utilizing Anthropic’s technology, a move typically reserved for organizations deemed hostile.
Emil Michael, acting as the Pentagon’s chief technology officer, emphasized the importance of diversification in partnerships, stating, “What we’ve learned…is that it’s irresponsible to be reliant on any one partner.” He highlighted that the new agreements encompass a combination of both closed and open-source AI models. A source familiar with the negotiations noted that Nvidia and Reflection would provide open-source options, which can be operated without ongoing licensing costs or vendor access. Such approaches afford the Pentagon greater operational flexibility, diminishing the potential for over-dependence on a single commercial entity.
Currently, Anthropic’s AI model, Claude, remains authorized for classified military use, despite the company’s ongoing legal battle against the Pentagon concerning these punitive measures. Tensions between the parties had appeared to lessen recently, as President Trump indicated he anticipated a healthier relationship following a meeting with Anthropic CEO Dario Amodei at the White House, with discussions described as “productive and constructive.”
The announcement outlined the incorporation of rival AI technologies as essential to the Trump administration’s initiative to create an “AI-first fighting force.” The agreements are expected to enhance data synthesis, improve situational awareness, and facilitate decision-making for military personnel. The Pentagon’s GenAI.mil platform, which is described as its official AI system, has reportedly been used by more than 1.3 million personnel, generating tens of millions of prompts and deploying hundreds of thousands of agents in just five months.
In a related development, over 600 Google employees recently called for the company to reject the Pentagon deal, reflecting ongoing internal dissent. This follows a successful employee movement in 2018 that compelled Google to withdraw from Project Maven—a military initiative aimed at incorporating AI into drone operations. In subsequent years, Google has pivoted its strategy, steadily re-engaging with military contracts and competing with other firms in the defense sector.
Project Maven has since transitioned to AI firm Palantir and evolved into a battlefield management system that significantly accelerates decision-making processes. Notably, during Operation Epic Fury against Iran, the platform reportedly enabled the processing of over 1,000 targets within a single day.
An AWS spokesperson expressed the company’s commitment to aiding the military and advancing modernization through AI. Other organizations involved in the agreements did not provide immediate comments on the matter.
