Korea’s Big Manufacturers Back Config, the Rising Star in Physical AI
South Korea and its tech giants are driving innovation in physical AI with Config, a startup backed by major manufacturers.
Admin User
Asia's push into physical artificial intelligence (AI) is gaining momentum thanks to the manufacturing prowess that has made it a global industrial powerhouse. In South Korea, Japan, China, and Taiwan, manufacturing remains a key driver of economic growth. Now, one startup is leveraging this strength: Config, a Seoul- and San Jose-based company building the data layer for robotics foundation models (RFMs).
Config has secured backing from the venture arms of South Korea's biggest manufacturers, including Samsung Venture Investment, Hyundai Motor’s ZER01NE Ventures, LG Technology Ventures, and SKT America. The oversubscribed $27 million seed round values Config at over $200 million, bringing its total raised to $35 million.
Founded in January 2025 by CEO Minjoon Seo, a former researcher at Meta and chief scientist at TwelveLabs, along with three co-founders from Waymo, Google, and Naver, Config is focused on providing data that robots need to learn and operate. Their goal is to make better data accessible for more useful robots.
Training large language models can be expensive due to the computing power required, but Config's challenge lies in physically collecting training data. Every piece of data requires a robot, a facility to run it, and people to operate it, making robotics AI more costly than software-only chatbots. Config wants to be the company that makes everyone else’s robot AI possible.
Config compares its role to TSMC, which manufactures for Apple, Nvidia, and AMD without competing with any of them. Similarly, Config aims to supply data for robotics without directly competing with other vendors.
The startup is already generating revenue through its customers, including large manufacturers, system integrators, and companies in agriculture and defense sectors. Its current data collection operation involves nearly 300 workers across Seoul and Hanoi, with over 100,000 hours of human motion data collected so far—more than 30 times the size of AgiBot World's open-source dataset.
Config's unique approach is in transforming the raw data before training begins to make it better suited for robots. CEO Minjoon Seo compares this process to language translation, saying that trying to teach Korean using only English materials is like expecting a model trained on one type of data to work seamlessly in another setting.
The funding will be used to scale its data operation toward 1 million hours of collected data, grow its enterprise platform business to $10 million in annual recurring revenue (ARR) by the end of 2027, and launch a cloud-based robot-as-a-service product that allows companies to run Config's foundation model without requiring onboard hardware.
