In a wind-swept conference hall on Spain’s Atlantic coast, diplomats last week assembled to discuss the most consequential military revolution since the atom bomb – AI.

Artificial intelligence — faster than missiles, cheaper than tanks and more pervasive than any previous weapon — is already reshaping warfare. Yet the defining feature of the gathering in A Coruña was not the principles agreed, but the reality exposed.

Europe spoke. The superpowers calculated.

At the Responsible AI in the Military Domain summit, a joint declaration calling for oversight, human accountability and safeguards for military AI was endorsed by just 35 countries. Missing from the signatures were the two states that matter most: the United States and China.

The symbolism was unmistakable. Europe attempted to write rules for a game in which it is no longer a serious player.

The declaration itself was mild, almost procedural. It emphasised that humans should remain responsible for lethal decisions, that AI systems require training and oversight, and that risks should be assessed before deployment. Sensible, civilised, and strategically irrelevant.

For while European governments debated ethical frameworks, Washington and Beijing were confronting a different question: not how to regulate AI warfare, but how fast it can be deployed.

Artificial intelligence is not another weapons platform; it is a universal military accelerator. It enhances targeting, command decisions, logistics, cyber-operations and electronic warfare simultaneously. The first power to integrate it deeply into operations does not merely gain an edge — it changes the nature of deterrence itself.

This is why both the United States and China declined to sign. Not because they reject safety, but because they cannot afford constraint. Each fears the other will move faster. Any nation that voluntarily limits its development risks falling irretrievably behind. Europe, uniquely, appears willing to test that theory in practice.

The uncomfortable truth is that Europe no longer sets the pace of military innovation. Its defence industry remains technically sophisticated, but structurally paralysed. Procurement cycles stretch across decades; major programmes require multinational compromise; political approval is fragmented among governments with conflicting priorities. By the time Europe fields a capability, the strategic environment has often already changed.

Artificial intelligence punishes delay more ruthlessly than any previous technology. Software evolves in months, not decades. Military advantage will belong not to the most advanced prototype but to the fastest iteration cycle — the side able to update operational algorithms weekly rather than annually.

Here Europe faces a cultural as much as institutional obstacle. Democracies debate; the European Union deliberates; bureaucracies harmonise. None of these processes function at the tempo of machine-learning development.

Washington, by contrast, can integrate Silicon Valley into defence planning almost overnight. Beijing can mobilise state, industry and academia under a single strategic directive. Europe convenes a working group.

The result is a familiar pattern: Europe becomes the regulatory power while others remain the strategic powers. Brussels drafts frameworks; others design systems. Europe refines compliance; others shape outcomes.

This might suffice in trade policy or digital privacy. It is dangerously inadequate in warfare.

The AI battlefield will not wait for consensus documents. Future conflicts may involve autonomous drone swarms identifying targets, cyber-defences responding in milliseconds, and predictive systems recommending pre-emptive moves. Decision time compresses to the point where hesitation becomes vulnerability.

The summit’s core fear — unintended escalation — arises precisely from this speed. When machines react faster than diplomats can communicate, accidents become strategic events. Yet managing such risk requires not merely moral authority but operational capability.

Europe possesses the first but increasingly lacks the second.

Several European countries, including Britain, France and Germany, endorsed the declaration in the hope of establishing early norms. But norms without enforcement rarely constrain rivals. Arms control historically succeeds only when the principal competitors accept mutual limits after achieving parity. Artificial intelligence has not yet reached that stage.

Instead, the continent risks occupying a peculiar position: morally earnest yet strategically marginal. It seeks to regulate technologies whose decisive development occurs elsewhere.

Even within the Western alliance, hesitation was visible. Reports indicated strained transatlantic relations left some states uncertain how far to align with Washington’s position. Europe could neither lead independently nor fully coordinate with its primary security guarantor. It became, in effect, a forum rather than a force.

This is not a judgement on European engineers or scientists, who remain among the world’s best. It is a judgement on political capacity. Rapid military adaptation requires decisive funding, unified command structures and tolerance for risk — precisely the qualities modern European governance struggles to provide.

The continent still thinks in terms of treaties; the new era operates in terms of iteration speed.

The deeper problem is psychological. Europe continues to view warfare as something that can be moderated through norms before capabilities emerge. History suggests the opposite. Regulation usually follows power, not precedes it. Nuclear arms control became possible only after both superpowers possessed thousands of warheads. AI warfare will likely follow the same trajectory.

Thus the A Coruña declaration may prove less the beginning of a regime than a record of an aspiration — a document capturing Europe’s desire to shape a future it lacks the means to command.

None of this renders the initiative pointless. Norms matter, and early frameworks may eventually guide behaviour once the United States and China see mutual advantage in restraint. But that moment will arrive through strategic equilibrium, not moral persuasion.

In the meantime, the decisive developments in military AI will occur not in conference halls but in laboratories linked directly to operational commands.

The sobering implication is that the AI arms race is already under way, and it is moving at machine speed. Europe hopes to slow it. America and China intend to win it.

History rarely waits for those who deliberate. In the coming decade, the measure of power may not be territory or fleet size, but the ability to adapt algorithms faster than an opponent. On that battlefield — silent, instantaneous and unforgiving — hesitation carries a heavier penalty than error.

Europe’s challenge is therefore stark. It must decide whether it wishes to remain a commentator on strategic change or again become a participant in it.

For the machines will not pause while committees agree. And neither will the nations programming them.

Main Image: Milrem Robotics – Own work 

What Ukraine’s “AI-driven air defence” is, and what it is not