Arthur Mensch stood in front of Brussels policymakers on Tuesday and said what no European official has been willing to say for years. The CEO of Mistral, France's leading AI company valued at over 11 billion euros, told the room that Europe's militaries run on American technology. That means a foreign government can shut them down.
His exact words: "If you don't have artificial intelligence in your systems, you actually don't have an army."
He is correct. And the implications go further than he stated.
Mistral AI valuation: over 11 billion euros. Europe's leading AI company, headquartered in Paris.
Verified
The Dependency Is the Vulnerability
Most of Europe's AI workloads run on infrastructure controlled by foreign providers, per Mistral's policy brief.
Verified
Most of Europe's AI workloads run on infrastructure controlled by American providers. Mistral's own policy paper, released alongside Mensch's remarks, states this plainly. Cloud services, semiconductor supply chains, foundation models: the entire stack traces back to companies headquartered in California and subject to American export controls, licensing restrictions, and political pressure.
Mensch posed the question that European defense ministries have avoided: "If these artificial intelligence systems are actually procured from foreign companies, then our militaries can be turned off. Do we want our military forces to be turned off because we have general political misalignment sometimes?"
The European Commission plans to release a major technological sovereignty package by end of May 2026, covering cloud, semiconductors, and data centers.
Verified
The word "sometimes" carries weight. The United States and Europe have disagreed on Iraq, Libya, Nord Stream sanctions, and now the conduct of operations in the Middle East. Washington pressured Anthropic last month over red lines the company set for military applications. The Trump administration wanted those guardrails removed. European governments watched and said nothing.
At Issue
The US government disputed Anthropic's restrictions on military AI use last month, signaling that American AI companies serve American strategic interests first.
Biased BipartisansReal-Time, Evidence-Based News Reports
Unlimited access to your personalized investigative reporter agent, sourcing real-time and verified reports on any topic. Your personalized news feed starts here.
Create Free AccountAI as a Deterrence Capability
Mensch compared artificial intelligence to nuclear weapons and deterrence. That comparison is more precise than most audiences realize. Nuclear deterrence works because a state controls its own arsenal. The launch authority, the delivery systems, the warheads: sovereign capability from top to bottom. No NATO member outsources its nuclear trigger to another country.
AI-enabled defense systems are heading toward the same level of strategic significance. Autonomous targeting, electronic warfare, logistics optimization, intelligence analysis: these functions will determine who wins the next conventional conflict. If the software running those systems answers to a foreign corporation subject to foreign law, the nation deploying them does not control its own defense.
The European Commission plans to release a technological sovereignty package by the end of May. That package will address cloud services, semiconductors, and data centers. Mistral wants the Commission to prioritize European-controlled AI infrastructure and use government procurement to keep critical workloads on European systems.
The Precedent That Should Alarm Every European Capital
Consider what happened with Anthropic. The US government disputed the company's restrictions on military use of its AI models. The dispute became public. The message to the industry was clear: American AI companies serve American strategic interests, and Washington will push back against any guardrails that interfere with those interests.
Biased BipartisansThink Further on BIPI.
Where seeking the truth is a journey, not a destination.
Learn moreNow apply that logic to a European ally running its command-and-control systems on American AI. In a scenario where Washington and Paris disagree on the use of force, who controls the software? The answer matters more than any treaty article.
Audrey Herblin-Stoop, Mistral's senior vice president for global affairs, said it is too early to address whether European governments will face pressure to lower AI safeguards in defense applications the way the Trump administration pressured Anthropic. She called for Europe to focus first on building its own ecosystem.
That answer is diplomatic. The strategic reality is less polite. Europe does not have time to build the ecosystem first and address the defense implications later. Both must happen now.
What Inaction Produces
The historical pattern repeats. Europe underinvested in defense for decades after the Cold War and then watched Russia invade Ukraine. Europe depended on Russian energy and then scrambled to find alternatives in the winter of 2022. Europe trusted American security guarantees and then watched Washington pivot toward bilateral deals and transactional alliances.
Mensch gave the first honest assessment from the industry. AI sovereignty is not a commercial preference. It is a defense requirement. Any European government that deploys foreign AI in military systems is handing an adversary a kill switch. The question is whether Brussels will treat this warning with the urgency it demands, or file it alongside a decade of ignored defense spending targets.
Weakness, as always, is provocative. Dependence on foreign AI for national defense is weakness dressed in a procurement contract.
