[Edaily Marketin YunJi Kim Reporter] A startup that trains robots by collecting footage of people washing dishes and folding laundry has secured the largest-ever seed investment in German history. Following generative AI, physical AI—which moves and performs tasks in real-world spaces—is emerging as the next-generation investment target, and venture capital (VC) interest is shifting from robot hardware to training data and on-site application capabilities.

According to the global investment banking (IB) industry on the 24th, Munich-based robotics startup MicroAGI recently secured $55 million (approximately 80.5 billion won) in seed funding from investors including Hummingbird Ventures, Northzone, LocalGlobal, and RedAlpine. This investment, secured approximately 10 months after the company’s founding, marks the largest seed round ever executed in the German startup ecosystem.

MicroAGI is a company that collects data on how people work in real-world spaces to support the on-site application of robots and AI models. It builds training datasets from the processes by which people pick up and move objects or handle tools, and uses this data to help robots adapt to the specific equipment and work environments of each factory. To this end, the company operates “Shift,” a dedicated data collection service. In New York, Shift provides free home cleaning services in exchange for filming workers performing tasks such as washing dishes, mopping floors, and folding laundry from a first-person perspective. Using cameras and sensor-equipped gloves, the service captures data on hand movements and work sequences.

While generative AI could utilize text, images, and videos accumulated on the internet as training data, there is a relative shortage of data that enables robots to learn how to work in the real world. Even for the same task—such as washing dishes—the shape and position of cups, the structure of the sink, and surrounding obstacles all vary, making it difficult for robots to adapt to real-world environments using only standardized laboratory data.

Venture capitalists have taken note of the potential for companies that secure this type of unstructured task data first to grow into key infrastructure providers in the future robotics market. They believe that companies supplying data to robots and AI models from multiple manufacturers—and handling on-site training and deployment—will be able to capture a broader market than companies that simply sell specific robots.

MicroAGI’s industry-focused business model was also cited as a key factor behind the investment. The company views industries with high volumes of repetitive tasks and labor shortages—such as automotive, logistics, and food production—as its primary markets. This strategy appears to address the fact that even as robot hardware performance improves rapidly, commercialization remains difficult without training robots to handle site-specific processes and exceptional situations.

Europe’s aging population and the shortage of manufacturing workers are also factors driving growth expectations. As demand for automation expands due to a shrinking workforce and rising wages—coupled with the trend of “reshoring,” which involves bringing manufacturing supply chains back within the region—the need for robotics adoption is growing.

Meanwhile, MicroAGI plans to use the funds from this investment to expand its computing infrastructure and data collection network, as well as to enter the U.S. market. Commenting on this, a foreign media outlet analyzed, “This is an example showing that the competitive focus in the robotics industry is shifting from hardware to data and operational software,” adding, “As the widespread adoption of robots gains momentum, the ability to secure diverse task data and apply it in real-world settings is likely to become a major barrier to entry.”