It wasn’t enough to experiment. Europe’s militaries needed results. Now they’re baking AI straight into their core defense capabilities. The shift is happening fast.
Germany and Ukraine kicked things into high gear on Monday. They launched “Brave Germany.” Think 5,000 drones. AI-enabled. Medium-range strikers. Jointly operated. This Berlin-Kyiv handshake isn’t isolated. It’s part of a wave of deals across the continent, stitching algorithms into everything from decision desks to weapon sights.
What’s actually out there?
Let’s clear up a misconception. Europe isn’t new to this. For ten years, they’ve used AI for the boring stuff. Logistics. Human resources. Maintenance schedules.
Laura Bruun at SIPRI notes the tech matured around 2015. That was when it stopped being a toy and became a priority.
“Very simple AI models… it’s faster if you take routeB than route A,” she says. Like Google Maps but for tanks.
But now? The money flows into two buckets. One, semi-autonomous weapons. Two, decision support.
With the weapons, humans stay in the loop. They still press the button. But for decision support, AI covers every task where a computer can help you choose a course of action in war. Battle management. Tactical planning. Roy Lindelauf, a professor at the Netherlands Defence Academy, says this is where the cash is burning.
The big players
Three names keep popping up. France. Germany. The UK.
They’ve signed huge contracts with private AI firms. Germany is moving fast on its Future Combat Air System. In 2023, their Ministry of Defence teamed up with Helsing AI to build the digital backbone of their next-gen fighter jets.
Wait. More.
Germany also signed Helsing to bolt AI onto the Eurofighter’s electronic warfare systems. Then another deal. €269 million for “kamikaze drones.” Loitering munitions for NATO stockpiles.
The UK has the Asgard programme. Announced for 2025 but already moving. A network combining sensors and strike tools. They want speed. They want lethality. And they’ve got a new best friend: Palantir. The US tech giant is pouring up to £1.5 into Britain. Help them harness AI? Yes.
France wants independence.
Sovereignty isn’t just a political word anymore. It’s a tech strategy.
They want military AI that doesn’t rely on America. In January, the government handed a framework deal to Mistral. A Paris-based shop competing with the American giants. Mistral’s models get access to French armed forces. It builds on 2025 cooperation pacts.
The EU itself is trying to catch up. The European Defence Fund just greenlit projects for sovereign large language models. AI for artillery. A private, sustainable system for member states.
Europe has plans. Lindelauf calls them well-thought-out. But execution? He’s worried. The bureaucracy moves too slow.
The Ukraine effect
Europe is learning by watching the fight in Ukraine. It’s the ultimate proving ground.
Ukraine built “Delta.” A digital battle command system. It eats data. From satellites. Radars. Trackers. The AI analyzes the noise. It tells officers where the enemy is. Where the friends are.
“It combines huge amounts of different data… the AI layer doing the analysis,” Lindelauf points out.
And the drones? Ukrainian forces use loitering munitions heavily. They aren’t fully autonomous killers. A commander says “strike.” The machine follows. But it’s automated navigation and identification.
Then there’s Palantir again. Working with Kyiv on “Brave1 Dataroom.” AI trained on combat data. Another system sorts air strike details. Another handles intelligence dumps.
The European Commission just launched STRATUS. Cyber defense for drone swarms. Who’s building part of it? A Ukrainian subcontractor. The tech will be tested on real battlefield mud. Not a lab.
There’s a darker edge emerging though. Bruun says Ukraine is testing full automation. What happens if a commander loses contact with a missile?
“I’ve read interviews… the human is a bottleneck,” she notes. If you can’t decide, you lose. So they automate the “finish the job” function. More automation equals resilience. Speed matters.
The lines are blurring. The decisions are getting faster. Who decides when the machine decides for itself?
