Korea's four biggest manufacturers usually compete for the same talent, customers, and supply chains. This week, they all wrote checks to the same startup.
Who Got The Money
Config closed an oversubscribed $27 million seed round at a valuation north of $200 million, with total money raised to date sitting at $35 million.
Samsung Venture Investment led the round, while Hyundai's ZER01NE Ventures, LG Tech Ventures, and SKT America (the venture arm of Korean telco SK) joined as strategic investors.
Pieter Abbeel, a UC Berkeley professor and Covariant AI co-founder, came in as an angel, alongside financial backers like Mirae Asset Ventures and Korea Development Bank.
The company was founded in January 2025 by CEO Minjoon Seo, a former Meta researcher and chief scientist at Twelve Labs, plus three co-founders with backgrounds at Waymo, Google, and Naver.
Why Korea's Industrial Giants Are Pooling Money
Config does not build robots. It builds the training data robots need to learn.
That sounds boring until you realize it is the bottleneck in physical AI, since training a chatbot is expensive but the raw material (text on the internet) is basically free.
Teaching a robot to fold a shirt or pack a box means physically recording a person doing it, hour after hour, in controlled studios.
Config has over 100,000 hours of human motion data on file, while the largest open-source data set has roughly 3,000 hours - a 30x gap on the most costly input in the industry.
If you want this kind of breakdown in your inbox every morning, Market Briefs is the five-minute daily read and it comes with a free investing masterclass when you sign up.
The TSMC Comparison
Config's pitch is that it can be the TSMC of robot AI, with TSMC being the Taiwanese chipmaker that builds chips for Apple, Nvidia, and AMD without competing with any of them.
Config wants to be that for robot data, selling to everyone and competing with no one.
The Korean manufacturers backing the round are doing exactly what you would do if you believed the pitch, which is to lock in early access.
CEO Minjoon Seo said the company's edge is that it transforms the data before training, instead of teaching robots with human motion footage and adapting later. He compared the work to teaching Korean with only English textbooks, where the materials need to be converted before they are useful.
Worth Noting
Config is already pulling in revenue, with customers including large manufacturers, system integrators, and companies in farming and defense.
The new money will fund three goals, which are scaling data collection in Vietnam and Seoul toward one million hours, hitting $10 million in yearly recurring revenue by the end of 2027, and launching a cloud-based Robot-as-a-Service product.
Korea's four biggest manufacturers do not write checks to the same startup unless they think it sits on something they need.
If you want a five-minute morning read on deals like this, Market Briefs is delivered weekday mornings and a free 45-minute investing course is in as a bonus.
