Samsung, Hyundai And LG Just Showed You Where Robotics Is Heading – And It’s All About Data

Three of the largest manufacturers in the world – Samsung, Hyundai and LG – have jointly backed a South Korean robotics data startup called Config, which has raised $27 million in a seed round led by Samsung Venture Investment at a valuation of over $200 million. The round also includes ZER01NE Ventures from Hyundai Motor Group, LG Technology Ventures, SKT America, GS Futures and Kakao Ventures, bringing Config’s total funding to around $35 million.

Config doesn’t build robots, it builds the data layer that trains them – collecting, cleaning and structuring motion and sensor data from industrial robots and physical AI systems so that manufacturers can use it to train their own robotic foundation models.

The company calls itself the “TSMC of robot data.” That analogy is doing a lot of work, and it’s a smart one. The positioning is intentional: like TSMC in semiconductors, Config aims to be a neutral foundry that serves everyone without competing with any of them.

Why The TSMC Analogy Works

TSMC manufactures chips for Apple, Nvidia and hundreds of others without building consumer devices itself. Its value comes from being the neutral, trusted infrastructure layer that everyone depends on but nobody controls. Config is attempting the same positioning in a different domain: it wants to be the company that Samsung, Hyundai and LG all rely on for robot training data – precisely because none of them want to depend on a competitor’s proprietary data stack.

That’s the rationale behind three rival manufacturers backing the same startup. By funding a neutral data infrastructure provider, each of them gets access to high-quality, standardised robot training data without ceding control to a competitor.

Config’s platform normalises sensor data across different robot makers into a common schema, which has the potential to become a de facto standard for the industry. The investors aren’t just backing a startup, they’re helping to build infrastructure they all need and none of them wants to build alone.

The Real Bet: Data Is The Bottleneck In Robotics

The Config raise spotlights where the true value in the robotics surge is accumulating. Building robots is hard and competitive, but data infrastructure for robotics is the real prize: it’s more scalable, more defensible, and perhaps more valuable than the hardware

The bottleneck in next-generation robotics AI is no longer just hardware – it’s high-quality, standardised, real-world data for training robotic foundation models. Config is going after that layer directly.

The roadmap reflects the ambition – Config plans to expand its Vietnam and Seoul-based data operations toward one million hours of motion data and scale up its enterprise platform to $10 million ARR by the end of 2027. It also intends to launch a cloud-based Robot-as-a-Service product that lets companies run its foundation models without needing onboard hardware.

That last move takes Config from data supplier to platform business, which is where the true advantage lies.

Strategic Takeaways for Physical AI Founders

The Config raise serves as a valuable benchmark for founders exploring infrastructure opportunities within physical AI.

This is a recurring trend within the software space: companies that built the data infrastructure, developer tooling and deployment layers around the last wave of AI, not the models themselves, captured enormous value. Physical AI is following the same trajectory, and the infrastructure layer is still largely being built.

The TSMC analogy also carries a lesson about positioning. Serving Samsung, Hyundai and LG concurrently without conflict of interest isn’t a constraint for Config – it’s a core strength. In markets where the major players are rivals, a neutral infrastructure provider that all of them need is one of the strongest competitive positions available.

If you’re a founder looking for emerging tech opportunities, keep this pattern in mind.