Fazlur Rahman
Published September 10, 2026 - Washington, D.C. The Pentagon has struck deals with OpenAI, Google, Microsoft, Amazon, Nvidia, Elon Musk's xAI, and the startup Reflection, allowing the agency to use their AI tools in classified settings, according to an announcement on May 1, 2026 (The Verge, May 1, 2026). The deals allow for the 'lawful operational use' of the AI systems, 'establishing the United States military as an AI-first fighting force.' At the same time, the Defense Department has left out Anthropic, which it previously used for handling classified information, after declaring the AI lab a supply-chain risk over a dispute about red lines around mass domestic surveillance and fully autonomous weapons.
Pentagon AI deal data last verified September 10, 2026 from The Verge (May 1, 2026) and Department of Defense public announcements.
The Pentagon struck classified AI deals on May 1, 2026 with OpenAI, Google, Microsoft, Amazon, Nvidia, xAI, and Reflection. Anthropic was excluded after the Defense Department declared it a supply-chain risk over its refusal to loosen red lines around surveillance and autonomous weapons. The deals establish the legal framework for commercial AI to be used in classified military environments. The Trump administration is taking a more aggressive posture on AI adoption than the Biden administration did.
The Seven AI Companies That Got Pentagon Access
The Pentagon deals cover seven AI companies across the AI stack: model providers (OpenAI, Google, xAI, Reflection), cloud and infrastructure providers (Microsoft, Amazon), and compute hardware provider (Nvidia). The deals allow for the 'lawful operational use' of the AI systems, 'establishing the United States military as an AI-first fighting force' (The Verge, May 1, 2026). The contracts with Nvidia and Reflection are new, while the others (OpenAI, xAI, Google, Microsoft, Amazon) had earlier deals that were formalized under this umbrella announcement.
The contracts are not direct procurement awards but rather access agreements that allow the AI systems to be used in classified Pentagon environments. The companies retain ownership of their AI models and intellectual property, and the Defense Department does not receive source code or model weights. The framework is intended to accelerate the Pentagon's adoption of commercial AI technology while maintaining the security controls required for classified environments.
Why Anthropic Was Excluded
Anthropic was excluded from the Pentagon AI deals because the Defense Department declared it a supply-chain risk after Anthropic refused to loosen 'red lines' around mass domestic surveillance and fully autonomous weapons for the agency (The Verge, May 1, 2026). Anthropic had a $200 million deal to handle classified materials for the Pentagon, but the dispute over red lines led to a ban on the AI startup's products from the federal government. Anthropic sued the federal government in response and won a temporary injunction, but the May 1, 2026 announcement formally excluded Anthropic from the broader Pentagon AI access framework.
Anthropic's red lines include: no use of Anthropic's models for fully autonomous weapons systems (i.e., weapons that select and engage targets without human oversight), no use for mass domestic surveillance of US persons, and no use for autonomous cyber-offense operations. The Defense Department wanted Anthropic to loosen these red lines as a condition of continued partnership, and Anthropic refused. The refusal was a strategic decision by Anthropic's leadership, who believed that loosening the red lines would damage the company's reputation with enterprise customers and the broader public.
The Paradox of Anthropic's Security Model
Emil Michael, the Defense Department's chief technology officer, told CNBC on May 1, 2026 that Anthropic is still a supply chain risk, but called its powerful security model Mythos a 'separate national security moment,' adding that 'we have to make sure that our networks are hardened up, because that model has capabilities that are particular to finding cyber vulnerabilities and patching them' (The Verge, May 1, 2026). The statement is notable for the tension it captures: the Defense Department views Anthropic's model as both a security risk (for refusing to loosen red lines) and a security asset (for its cybersecurity capabilities).
The result is a paradoxical relationship where the Pentagon publicly excludes Anthropic while privately acknowledging the value of its technology. The Mythos model is a specialized AI security model that Anthropic developed for vulnerability discovery and patching, and the Defense Department's CTO is essentially admitting that the model has national-security value even as the broader agency is excluding Anthropic. The situation is likely to be revisited as the cybersecurity threat landscape evolves, particularly as state-sponsored AI cyber attacks become more sophisticated.
What the Deals Mean for the AI Companies
For OpenAI, Google, and Nvidia, the Pentagon deal is a major validation of their AI technology for the most demanding security use cases, and a significant revenue opportunity as the Defense Department ramps up AI spending through the rest of 2026 and 2027. For OpenAI, the deal formalizes the company's pivot from consumer-focused AI to enterprise and government AI, which has been a strategic shift since the November 2023 board drama.
For Google, the deal is a return to the defense market after years of employee protests against the company's involvement in military AI projects. Project Maven, the Pentagon's flagship AI initiative, was the focus of significant employee protest at Google in 2018, leading to the company's withdrawal from the project. The May 2026 deal signals that Google has decided that the strategic value of Pentagon AI work outweighs the employee-relations risk, particularly as the competitive AI landscape has intensified.
For Nvidia, the deal includes both the AI compute hardware and the company's own AI software stack, which is increasingly competitive with the model providers' offerings. Nvidia's DGX systems and the Nvidia AI Enterprise software suite are now used across the AI industry, and the Pentagon deal extends that reach into the defense market.
Strategic Implications for the AI Industry
The Pentagon deals also signal that the Trump administration is taking a more aggressive posture on AI adoption in the federal government than the Biden administration did. The Biden administration issued the Executive Order on AI in October 2023, which established oversight requirements for federal AI use. The Trump administration's approach is more permissive, focusing on accelerating AI adoption rather than establishing guardrails.
For the AI industry, the Pentagon deals represent a major source of new revenue and a validation of the strategic importance of commercial AI for national security. The deals also create a competitive advantage for the companies that have access, particularly in the classified-AI market where the barriers to entry are high (security clearances, classified cloud infrastructure, etc.). The exclusion of Anthropic from the deals is a strategic setback for the company and may be a factor in the IPO pricing when Anthropic goes public in 2027.
Verify current Pentagon AI deal status on the official Department of Defense press release portal at defense.gov and the individual company press releases.
