Config Secures $27M Seed at $200M+ Valuation to Build Data Infrastructure for Robotics AI, Backed by Samsung Ventures

Config, a startup building a foundational data layer for robotics AI, has raised $27 million in seed funding at a valuation exceeding $200 million, led by Samsung Venture Investment. The round underscores surging investor appetite for the infrastructure powering the next wave of physical artificial intelligence in Asia.

The San Francisco-based company, which operates primarily from South Korea and Silicon Valley, specializes in curating and labeling real-world sensor data—including vision, touch, and force feedback—to train what it calls "robotics foundation models." These models aim to give robots human-like perception and dexterity, unlocking new capabilities in manufacturing, logistics, and home assistance.

"What we’re doing is analogous to what companies like Scale AI did for language models, but for the physical world," said Alex Kim, CEO of Config, in an exclusive statement. "Our data layer makes it possible for robots to learn from billions of real interactions rather than just simulated ones."

The investment comes as Asian manufacturing hubs double down on automation and embodied AI. According to industry analysts, the region’s deep supply chain expertise and mass production capacity create a natural home for robotics innovation.

Background

Config was founded in 2023 by a team of former researchers from Korea Advanced Institute of Science and Technology (KAIST) and robotics engineers who previously worked at Samsung and Hyundai. The company’s core platform ingests raw sensor streams from industrial robots and human teleoperation, then annotates them for machine learning training.

Robotics foundation models are large AI models trained on diverse robot data, much like GPT or BERT for language. They promise to reduce the time and cost of developing custom robot skills by enabling "zero-shot" or few-shot learning across different robot hardware. However, they require massive amounts of high-quality, multimodal data—a bottleneck Config aims to solve.

"Without a robust data infrastructure, even the best algorithms are useless in real factories," noted Dr. Soo-Young Lee, a robotics professor at KAIST and informal advisor to Config. "This funding will help build that pipeline at scale."

What This Means

The $27 million seed round at a $200 million-plus valuation signals that venture investors see data as the new moat in robotics. Samsung Venture Investment’s lead also points to conglomerates betting on open infrastructure rather than vertical integration alone.

Industry watchers say this capital will accelerate Config’s plan to expand its data annotation workforce in Southeast Asia and forge partnerships with robot manufacturers. If successful, Config could emerge as a critical middleware layer—much like what Nvidia is for AI compute or what Databricks is for big data—in the physical AI stack.

"We are on the cusp of a robotics data explosion," said Kim. "This investment gives us the runway to become the standard data layer for every robot that needs to learn the real world."