Pentagon CTO opposes government stakes in AI companies, warns against regulation

The Pentagon’s top technologist wants the US government to keep its hands off AI companies, both financially and regulatorily. Emil Michael, who serves as the Department of Defense’s Chief Technology Officer and Under Secretary of Defense for Research and Engineering, went on CNBC to argue that Washington should neither acquire equity stakes in AI firms nor pile on new rules that could slow the industry down.

The anti-regulation case from inside the Pentagon

Michael’s core argument is that AI companies partnering too closely with politicians on regulation is a recipe for trouble. The implication: regulatory frameworks tend to calcify around whoever’s already winning, locking out the next generation of competitors and slowing down the kind of rapid iteration that makes AI useful in the first place.

The Pentagon has been on an aggressive tear integrating commercial AI tools into its classified networks, working with OpenAI, Google, xAI, Nvidia, and Microsoft, among others.

The stance also comes with a notable exception that proves the rule. The DoD has made equity investments in certain suppliers where strategic supply chain concerns demanded it, including a 15% stake in MP Materials valued at $400 million. MP Materials processes rare earth elements, the kind of critical minerals where the US has genuine supply vulnerability. AI companies, in Michael’s view, don’t present the same existential dependency problem.

An AI-first military takes shape

Michael’s comments sit against a backdrop of sweeping procurement reforms at the Pentagon under Defense Secretary Hegseth. The department has consolidated its various innovation offices under Michael’s authority, creating a streamlined pipeline that prioritizes six critical technology areas, with applied AI sitting at the top of the list.

GenAI.mil, the Pentagon’s internal generative AI platform, has scaled to 1.5 million personnel as of mid-2026. The entire active-duty US military numbers roughly 1.3 million people, meaning the platform has clearly spread beyond warfighters into the civilian and contractor workforce that keeps the defense bureaucracy running.

Anthropic has effectively been sidelined from Pentagon partnerships. The disputes reportedly center on Anthropic’s guardrails, the safety constraints it places on its models.

What this means for the AI industry

The vendor consolidation happening under Michael’s watch is creating a tiered marketplace. At the top sit the companies with established classified network access: OpenAI, Google, Microsoft, Nvidia, and Elon Musk’s xAI.

Michael’s vision removes one potential funding source, government equity, but opens up a larger revenue opportunity through contracts. The tradeoff favors companies with mature products over those still looking for patient capital.

The 1.5 million user figure for GenAI.mil creates a significant procurement dynamic. At that scale, switching costs become enormous, and whichever companies power the platform’s backend are building the kind of institutional lock-in that defense contractors have historically used to generate decades of revenue.