Why a Red Bull Engineer Just Raised $55M to Train Robots With Real-World Data

A Formula One aerodynamicist walked away from Red Bull Racing’s dominance to solve a problem holding back the entire robotics industry: robots can’t learn from real-world chaos the way humans do. Four days ago, his startup raised $55M—Germany’s largest seed round ever—by proving that the fastest path to factory automation isn’t building new hardware or new AI models, but filling the data gap between them.

The Founder Story: Why Elite Talent Leaves Winning Teams

Bercan Kilic spent years as an aerodynamicist at Red Bull Racing, optimizing thousandths of a second on the most advanced machinery in motorsport. By every conventional measure, he had the dream job. Then he quit.

“The engineering was magnificent,” he said. “The point of it was thin.”

He wasn’t burned out or chasing a quick exit. He was looking for a problem that mattered. In September 2025, Kilic co-founded microagi with a WhatsApp entrepreneur and an Alan Turing researcher—a deliberate mix of aerospace precision, consumer-scale growth, and AI research expertise. Within ten months, they’d raised $55 million.

This isn’t just a funding story. It’s a template: elite talent from different worlds spotting a quantifiable bottleneck, moving fast, getting real customers, then raising capital on traction rather than promises. The capital signal matters, but the founder pedigree matters more—it tells you the problem is real enough to pull serious people away from prestigious positions.

The Problem: A 100,000-Year Data Gap

Here’s the bottleneck nobody talks about: we have world-class robot hardware and powerful AI models, but robots can’t generalize the way humans do.

A human learns a new factory job in a week. An existing robot model might never learn it. Why? Because robots train on lab footage—controlled lighting, perfect conditions, the same task repeated thousands of times. Real factories are chaos: different materials, unexpected obstacles, variable lighting, human unpredictability.

Ken Goldberg, a leading UC Berkeley roboticist, quantified the gap: robots need approximately 100,000 years of real-world training footage to match human generalization ability. That’s not hyperbole. That’s the actual infrastructure deficit.

Most robotics funding has gone to two places: new hardware (Walden Robotics raised $300M for legged construction robots; All3 is building robots for dangerous jobs) or new models (the foundation model arms race). But the layer in between—real-world footage that teaches existing models to work in existing factories—was almost entirely ignored.

Microagi saw the opening.

The Solution: Shift (Viral Consumer Layer) + B2B (Real Business)

Microagi’s approach has two layers, and the first one went viral.

Shift is the consumer-facing product: microagi offers free apartment cleanings, laundry services, and private chef work in exchange for filming the workers. You get a clean apartment; they get hours of real-world footage of someone mopping, folding, washing dishes, navigating a real space with real obstacles.

It sounds almost too simple. But it worked. Operating in 15 countries, they’ve recruited over 20,000 people to record themselves. When they launched in New York with free cleanings, it went viral. They’re now running private chef services in San Francisco—each interaction is training data.

But that’s the lead magnet. The real business is B2B.

Microagi sends workers into customer factories wearing camera rigs and sensor-equipped gloves. They film real workers doing real jobs—assembling parts, moving materials, operating machinery—in real factory conditions. Then microagi feeds that footage into existing robot models to teach them that specific task in that specific environment.

Five companies are already on the platform. One is preparing live robot deployment. Customers span automotive, logistics, and food production.

Kilic framed it clearly: “We provide the labs with data, they provide us with models, and then we layer on proprietary data to make our customers happy.”

They’re not replacing hardware companies. They’re not competing with model labs. They’re filling the gap between them—and that gap is where the actual bottleneck lives.

Why $55M Validates a Contrarian Thesis

The funding round itself is historic: $55M for a seed round led by Hummingbird Ventures, with participation from Northzone, LocalGlobe, Village Global, and redalpine. For a German startup, it’s the largest seed ever in a single funding event.

But the real signal isn’t the size. It’s what it says about where the robotics industry is heading.

China installed 295,000 factory robots in 2024—54% of the global total. The US installed 34,200. Europe and the US are facing worker shortages. Automation isn’t coming; it’s accelerating. The question isn’t whether factories will automate, but how fast and who wins.

Most robotics capital has assumed the winner would be the company with the best hardware or the best model. Microagi is betting on the software and data layer between them—a contrarian thesis that just attracted elite capital. That’s the validation: not just the dollars, but the fact that serious investors believe the fastest path to scale in robotics is optimizing the layer between existing robots and existing models, not building new ones.

The European Deep Tech Signal

Microagi’s funding is also a statement about where hard-tech innovation is happening. This is Munich, not Silicon Valley. This is a startup with a German founder, German investors, and a German base raising the country’s largest seed round.

For years, deep tech—hard problems requiring serious engineering—was assumed to be a US monopoly. Europe had talent and capital, but the narrative was that serious founders went to the Bay Area. Microagi (along with other recent European deep-tech wins like [RELATED: European robotics and hard tech funding trends]) suggests that narrative is shifting.

The playbook is clear: elite talent from different worlds, a real quantifiable problem, real customers, and then capital follows. That playbook works in Munich just as well as in Palo Alto.

FAQ

Q: How is microagi different from Scale AI or other data companies? A: Scale AI focuses on general data labeling and annotation. Microagi is specialized in robotics—they’re not just collecting footage, they’re collecting footage with sensor-equipped gloves and specialized rigs designed to teach robots specific tasks in specific factories. They’re also running the consumer-facing Shift layer, which gives them direct access to real-world movement data at scale.

Q: Is the Shift product just a marketing gimmick? A: No. It’s a lead magnet that also generates real training data. Every free cleaning or chef service is an hour of real-world footage of human movement in real environments. That data is valuable to the B2B business. But the real revenue comes from the factory side—charging customers for proprietary training data and integration.

Q: Why does the robotics industry need this? Don’t existing models already work? A: Existing models work in controlled lab conditions. They fail in real factories because of variability—different materials, lighting, obstacles, human unpredictability. The 100,000-year data gap is real. Factories need models trained on their specific conditions, and that requires real-world footage that labs don’t have.

Q: Is $55M enough to scale this? A: For a seed round, $55M is substantial. It gives them runway to expand Shift to more countries, grow their B2B customer base, and build proprietary training pipelines. But scaling robotics is capital-intensive. They’ll likely raise a larger Series A once they demonstrate clear revenue traction from the factory side.

The Takeaway: The Data Layer Is the Bottleneck

The next decade of factory automation won’t be won by the company with the best hardware or the best model. It’ll be won by whoever solves the data problem fastest—whoever can reliably generate real-world training footage at scale and integrate it into existing systems.

Microagi just proved that thesis is fundable. A Red Bull engineer left a winning team because he saw a thin point and decided to make it thick. Four days ago, $55M worth of serious capital agreed with him.