Forterra just pulled back the curtain on something that’s been running quietly for nine months. The U.S. autonomous vehicle builder revealed that more than 100 of its self-driving ATVs have been operating in Ukrainian conflict zones since last October, according to TechCrunch AI. The company believes it’s the largest deployment of autonomous ground vehicles in combat by any U.S. defense tech firm.
This is significant because ground robots have lived in the shadow of aerial drones, which grabbed most of the attention in this war. But the drones changed the math on the ground. Constant surveillance from above created wide no-go zones where any human movement can bring death in seconds.
“There’s nowhere to hide,” Sergeant Major Corey Wilkens, who leads a U.S. Army program developing autonomous vehicles and tactics, told TechCrunch AI. Soldiers become “very, very vulnerable” to FPV drones, dropped munitions, artillery, and mortar fire. Moving supplies or wounded soldiers by hand is a death sentence in those zones. That’s the problem Forterra’s machines are built to solve.
What the numbers show
Forterra’s Lancer vehicles are based on Polaris ATVs, fitted with a custom sensor and compute stack. Since arriving in Ukraine, they’ve racked up a real combat record, per TechCrunch AI:
- More than 2,500 miles driven
- Over 1,100 missions completed
- 777,440 pounds of cargo hauled
- 52 casualty evacuations
Some have been lost, mostly when they got stuck in deep mud and Russian forces picked them off. But the value is clear to the people using them.
Ukraine already builds its own uncrewed ground vehicles, but those are typically battery-powered and top out around 250 kilograms of cargo. The gas-powered Lancer carries 750 kilograms. One Ukrainian soldier, kept anonymous for security, put it bluntly to TechCrunch AI: this logistics UGV is “the most important UGV in Ukraine,” adding, “we are dying to get more.”
The Starlink lesson
It didn’t start smooth. Ukrainian forces have had mixed results with Western contractors, and Forterra’s gear at first felt too tuned for high-end U.S. Army needs. The fix that changed everything was practical: bolting on a Starlink antenna. That one modification turned the vehicle into a genuine asset.
What stands out here is how much of the “autonomy” is still human-driven. Ukrainian soldiers mostly teleoperate the vehicles in combat. Part of that is because the machines are too valuable to lose. Part of it is that the software still can’t handle a live enemy.
“We actually need to be able to respond to the enemy threats, live, while it’s in front of the enemy, which the autonomy doesn’t know how to do yet,” the Ukrainian soldier explained to TechCrunch AI. The vehicles can navigate rough terrain on their own. They can’t yet spot an unexpected ambush and react.
Why it matters for AI
Forterra has been working on autonomous vehicles for 20 years. Now it’s trying to fuse classic self-driving-car algorithms with newer generative AI that lets machines react to novel situations. The bottleneck is familiar to anyone tracking AI: data. Chief growth officer Scott Sanders, a former Marine officer, told TechCrunch AI that a lot of combat behavior “isn’t available in an open source model because they’re not things that humans do,” like navigating a minefield or operating a weapon system.
The field is heating up. Scout AI raised $100 million this year to build foundation models for military platforms. Field AI and Overland AI are running UGV trials with the U.S. military. Forterra, backed by more than $500 million in venture funding, has turned its Ukraine deployment into a proving ground for future national security contracts.
The practical takeaway: ground autonomy has crossed from concept to combat, but full autonomy hasn’t. “Ground autonomy is achievable now and we’ve seen it,” Wilkens said. Expect the next phase to be a race on two fronts, making the machines smarter and making them cheaper. The Ukrainians already issued that second challenge directly.
You can find the full report at the original source.