Asia's burgeoning interest in physical AI is being propelled by the same manufacturing capabilities that established the region as a global industrial leader. Countries like South Korea, Japan, China, and Taiwan continue to prioritize large-scale production and optimized supply chains, which are now influencing the adoption of artificial intelligence and directing investment.
In this context, Config, a startup based in Seoul and San Jose, has made headlines by securing significant investment from South Korea's largest manufacturers. This startup is focused on creating the essential data layer for robotic foundation models (RFMs), which are critical for advancing robotics technology.
Leading the charge is Samsung Venture Investment, which spearheaded an oversubscribed $27 million seed funding round that valued Config at over $200 million, raising its total funding to $34 million. Other notable investors include Hyundai Motor's ZER01NE Ventures, LG Tech Ventures, SKT America, and angel investor Pieter Abbeel, a co-founder of Covariant AI and a professor at UC Berkeley.
Founded in January 2025 by CEO Minjoon Seo, a former Meta researcher, Config's team includes experts from prominent companies like Waymo and Google. Rather than building robots, Config aims to provide the data necessary for these machines to learn and function effectively, believing that superior data is crucial for enhancing robot utility.
Training large language models is resource-intensive, but the challenge of teaching robots to move is even greater, as it requires extensive physical data collection. Seo notes that the costs associated with gathering and labeling this data can escalate quickly as companies strive for more advanced robotic capabilities.
Config envisions itself as the backbone of robotic AI development, much like TSMC has become for semiconductor manufacturing. By supplying essential data, Config is positioning itself to support manufacturers looking to develop proprietary robotic solutions rather than relying solely on external vendors.
Currently, Config is generating revenue and serving a diverse clientele, including major manufacturers and companies in sectors such as agriculture and defense. The startup's approach is gaining traction, with increasing interest from large manufacturers eager to create their own robotic AI systems.
Config's unique method involves recording human movements in controlled environments, amassing over 100,000 hours of data--far surpassing the largest comparable datasets available. This data is then transformed to better align with the movements and interactions of robots, which Seo likens to translating languages. The goal is to convert the data to suit the model, enhancing its applicability in real-world scenarios.
The newly acquired funding will focus on expanding data operations, aiming to reach one million hours of collected data, increasing its enterprise platform business to $10 million in annual recurring revenue by 2026, and launching a cloud-based Robot-as-a-Service product that allows companies to utilize Config's models without needing onboard hardware.
