Vydar wants to make GPS-denied navigation a European stronghold
Delft startup unveils sovereign edge-AI module that keeps drones flying when GPS collapses.
Published on December 25, 2025
Vydar's CFO Martijn Crijnen at Blue Magic Netherlands © Nadia ten Wolde
Bart, co-founder of Media52 and Professor of Journalism oversees IO+, events, and Laio. A journalist at heart, he keeps writing as many stories as possible.
When Martijn Crijnen walked on stage at Blue Magic Netherlands, he didn’t need slides to make his point. “On today’s battlefield, GPS is gone,” the Vydar CFO opened bluntly. In Ukraine, that’s not a metaphor: pervasive jamming means only one in five drones reach their targets. “They use over 10,000 drones every single week. A 20% success rate is catastrophic.”
The Delft-based startup believes it has the answer, and it fits into a 50×50 mm box.
Blue Magic Netherlands
Read our stories about this event, focused on innovation in the defense industry.
From GPS to computer vision: how Vydar redefines navigation
Vydar’s module replaces satellite navigation with onboard intelligence. “We use AI to look through a camera, recognize the ground beneath, and match those features with pre-processed satellite maps stored offline,” Crijnen explained. That match, fine-tuned between every frame and fused with IMU data, lets the drone compute an accurate coordinate even in a complete GPS blackout.
On-screen, a red line showed the ground-truth GPS path. A grey line showed Vydar’s estimation. They overlapped almost perfectly. Tested over a 1,000 km² map, the system achieves 20-meter accuracy without any satellite link. Accuracy improves further with sensor fusion, and the module works across seasons, as shown in Vydar’s winter–summer test imagery.
“We believe we can increase mission success from 20% back up to 90%,” Crijnen said.
European hardware, European autonomy
The geopolitical argument is just as important as the technical one. Virtually every competitor depends on U.S.-designed, Asia-manufactured AI chips, most notably NVIDIA Jetson. That creates cost issues, strategic vulnerability, and growing political restrictions. A Reuters clipping in Vydar’s deck bluntly stated the risk: US embeds trackers in AI chip shipments.
Vydar's CFO Martijn Crijnen at Blue Magic Netherlands © Nadia ten Wolde
Vydar chose a harder path: build the stack itself.
Their system is now 100% manufactured in Europe and weighs 30 grams, vs 176 grams for Jetson-based competitors. It has a 3-watt power consumption, an 88% efficiency improvement.
“We don’t make our own AI chips,” Crijnen noted, “but because the hardware is purpose-built, we can easily integrate future European-made chips. That flexibility is critical.”
A platform, not a single product
Vydar’s hardware is modular to support multiple defense applications. Think of terminal guidance and drone interception, military-grade FPV night vision using low-noise optical imaging, and distributed sensor networks enabling real-time situational awareness.
Future iterations go even further, Crijnen explained. The company is already developing an AI-to-FPGA platform that enables any AI model to run on low-cost, EU-manufactured reconfigurable chips. If they succeed, this could remove Europe’s dependency on foreign GPU fabs altogether, a recurring theme in Vydar’s strategy.
From Ukraine test flights to industrial scale
Vydar’s progress since its 2024 founding has been fast:
- Winner of the European Space for Defence hackathon
- Partnerships with drone manufacturers in the Netherlands and Ukraine
- A memorandum of cooperation with the Ukrainian state research institute
- A dedicated production line able to produce 100,000 modules every six weeks
The team has been flight-testing in the Netherlands and France and will conduct extensive trials in Ukraine before the end of the year.
Their business model is straightforward: sell modules to system integrators, who then supply ministries of defense. The Ukrainian market alone could exceed €300 million annually, with a global addressable market of €1.2 billion according to Crijnen.
“European engineering can create sovereign, scalable tech”
Crijnen closed his talk with a quiet confidence unusual for such a young company: “The problem of GPS-denied navigation is not just an aerospace challenge. It’s electrical engineering, software engineering, and brutal wartime economics. Every interceptor must be cheaper than what it destroys. Every navigation module must scale.”
If Vydar’s early results hold up under battlefield conditions, the Netherlands may soon be home to a deep-tech company proving that sovereign, low-cost AI hardware isn’t just desirable, but achievable.