← Back to feed

On May 28, Mistral CEO Arthur Mensch gave the most direct corporate rebuttal yet to Pope Leo XIV’s first encyclical, published three days earlier. The Vatican’s 42,300-word Magnifica Humanitas called for AI to be “disarmed” from military use, for binding international rules on autonomous weapons, and for human oversight of all lethal decisions. Mensch, speaking at a company event in Paris, dismissed it as an argument Europe cannot act on unilaterally.

“We’re all for peace, but if you look at our rivals and adversaries in the world, they’re using artificial intelligence. As long as we have adversaries that are threatening, and they are threatening, we do need to have our own capabilities.”

The Encyclical

Pope Leo XIV framed Magnifica Humanitas as a Rerum Novarum moment — the 1891 encyclical that set Catholic social teaching on industrial labour. He explicitly rejected traditional just-war theory as “outdated” and called for three binding requirements on any autonomous-weapons deployment: traceability of decisions, meaningful human control over lethal action, and international rules to slow the arms race.

The document received formal expressions of respect from OpenAI, Google, and Microsoft. Anthropic co-founder Christopher Olah appeared at the launch, lending Silicon Valley validation to the text. The European Commission welcomed it the same day.

Mensch’s response is a structural defence of the business line, not a theological argument. The distinction Pope Leo draws — self-defence in the strictest sense — and the position Mensch articulates are not strictly in contradiction. Both accept the legitimacy of defensive capability. The divergence is on what “defensive” requires in 2026 when adversaries are deploying AI across battlefield systems.

The Infrastructure Announcement

Alongside the rebuttal, Mistral announced a new data centre in Les Ulis, France — 10 megawatts, opening in H2 2026. The facility is the first step in a €4 billion investment strategy with specific milestones:

• 200 MW by end of 2027
• 1 GW by 2030

Existing capacity sits in Sweden and France. Mistral said the new infrastructure will serve its own inference needs and be rented to other AI labs — a positioning play toward becoming European compute infrastructure, not just a model provider.

The 1GW target is Mensch’s publicly stated floor. In testimony before French MPs in May, he argued Europe needed at least 40GW nationally and 400GW continentally within five years to avoid a €1 trillion annual AI services trade deficit.

Context: Mistral’s Defence Portfolio

The Pope’s encyclical is not targeting a hypothetical posture. Mistral has been building defence-AI capabilities since at least February 2025, when it announced a partnership with Helsing — a European defence AI company that has deployed systems in Eurofighter combat jets, battlefield simulations, and Ukraine drone operations. Mistral also works directly with the French Ministry of Armed Forces.

The company is valued at €11.7 billion, targeting €1 billion in annual revenue by end of 2026, and investing approximately €1 billion per year in R&D. American investors hold less than 30% of its capital. Mensch has consistently said a US acquisition would represent mission failure.

The Policy Tension

Brussels is moving toward enforceable AI-warfare frameworks but has not yet codified the binding restrictions Magnifica Humanitas calls for. European member states are simultaneously expanding defence-AI procurement budgets. Mensch’s public stance bets on the procurement side of that tension winning out.

The Vatican has no regulatory authority over AI development. What Magnifica Humanitas provides is a moral vocabulary legislators can use. Mensch’s rebuttal acknowledges, by its existence, that this vocabulary now has commercial stakes.