South Korea's Four Biggest Conglomerates Back Config, a Robot Data Startup Positioning Itself as the TSMC of the Robotics Industry
Samsung, Hyundai, LG, and SK jointly invested in Config's $27M seed round to build neutral training-data infrastructure for robotic foundation models.
Overview
Config, a Seoul- and San Jose-based startup founded in January 2025, has raised a $27 million seed round led by Samsung Venture Investment at a valuation above $200 million, bringing its total funding to $35 million. The round was oversubscribed and drew in all four of South Korea’s largest industrial conglomerates: Samsung, Hyundai Motor, LG, and SK — an unusual alignment of rivals around a single early-stage company. Config’s pitch is that it can become the robotics industry’s equivalent of TSMC: a neutral data infrastructure provider that trains the AI brains of robots without competing with any of its customers.
What We Know
Config was founded by CEO Minjoon Seo, a former Meta researcher and ex-chief scientist at TwelveLabs, alongside three co-founders with backgrounds at Waymo, Google, and Naver. COO Jack Bang is also a co-founder. The company currently employs nearly 300 people across operations in Seoul and Hanoi.
At the core of Config’s business is the collection and transformation of human motion data for use in training robotic foundation models. The company has accumulated over 100,000 hours of human motion data — approximately 30 times larger than AgiBot World, the largest comparable open-source dataset at roughly 3,000 hours. That gap in data scale is central to Config’s competitive argument.
The technical differentiation lies in how Config processes that data. According to Seo, the company’s method is to transform the data before training rather than train models on raw human motion footage and attempt to adapt them afterward. “The data must be converted, not the model. This conversion technology is Config’s core technical differentiator,” Seo told TechCrunch. Seo described the challenge as analogous to attempting to teach Korean using only English-language textbooks: the materials must be converted before they become useful.
Config normalizes sensor data from different robot hardware manufacturers into a standardized schema, potentially positioning itself as a cross-industry data standard — a role no individual robot maker could credibly fill without triggering competitive concerns among customers.
The investor lineup reflects that dynamic directly. Samsung, Hyundai, and LG all compete in hardware and manufacturing. They also all backed Config, alongside SKT America (the U.S. arm of SK Telecom’s venture unit). As TechRound noted, none of them want to depend on a competitor’s proprietary data stack. Additional investors include Mirae Asset Ventures, Korea Development Bank, GS Futures, Kakao Ventures, Z Ventures, and angel investor Pieter Abbeel, the UC Berkeley professor and Covariant co-founder.
Config is already generating revenue. COO Jack Bang confirmed the company’s paying customers span large manufacturers, system integrators, and companies in the agriculture and defense sectors. The new funding will go toward three priorities: scaling data collection toward one million hours, growing its enterprise platform business to $10 million in annual recurring revenue by the end of 2027, and launching a cloud-based robot-as-a-service product that allows companies to run Config’s foundation models without requiring onboard hardware.
What We Don’t Know
Config has not disclosed which specific manufacturers or system integrators it serves, or what per-unit or per-hour pricing looks like for its data products. It is also unclear how the company collects its human motion data — specifically, whether it relies on motion-capture studios, factory floor teleoperation, or other methods — and how representative that data is across the range of tasks humanoid robots will need to perform in real deployments.
The TSMC analogy is evocative but carries limits. TSMC’s neutrality is enforced by the enormous capital cost of semiconductor fabs, which makes independent manufacturing impractical for its customers. Whether a similar structural moat will exist for robot training data — a resource that multiple well-funded competitors such as Physical Intelligence, Generalist AI, and Skild AI are also pursuing — remains an open question.
Config has not disclosed a timeline for its cloud robot-as-a-service product or confirmed which foundation model the service will be built around.
Analysis
The composition of Config’s investor syndicate is arguably as notable as the funding amount. Persuading Samsung, Hyundai, LG, and SK — fierce competitors in hardware, automotive, and enterprise technology — to jointly back the same data infrastructure company at the seed stage is a signal that all four see high-quality training data, not hardware iteration, as the primary constraint on their ability to deploy capable robots at scale. That consensus, if it holds, could give Config a pathway to becoming an industry-wide standard for robotic motion data — a position that would be difficult for any single manufacturer to dislodge.
The 30-to-1 data advantage over the largest open-source alternative is meaningful today, but the field is moving rapidly. Whether Config can extend that lead fast enough to establish defensible network effects before better-funded rivals close the gap will determine whether the TSMC analogy proves prescient or premature.