Samsung and Hyundai Lead a Big Bet on Robot AI Infrastructure
Config has raised USD 27 million (approx. RM124 million) in seed funding to scale its robotic foundation models and AI data infrastructure, in a round led by Samsung Venture Investment. The deal reportedly values the startup at over USD 200 million (approx. RM919 million) and includes participation from major manufacturing-linked investors such as Hyundai Motor ZER01NE Ventures, LG Tech Ventures, SKT America, and others. For Samsung and Hyundai, this Samsung robotics investment marks a strategic move beyond hardware into the data and intelligence layer that will power future fleets of robots in factories, logistics hubs, and fields. Rather than building yet another robot platform, Config is positioning itself as a shared robot AI data platform on which OEMs and integrators can train their own models, signalling a shift toward proprietary, differentiated AI capabilities in robotics.
Why Robotic Foundation Models Need Their Own Data Stack
Config’s founding team, which includes alumni from Meta, Twelve Labs, Waymo, Google, and Naver, is focused on the hardest part of robot learning: collecting and converting the physical-world data that robotic foundation models need. CEO Minjoon Seo argues that investment requirements for robotics AI far exceed those of software chatbots because robots need data gathered by real machines operating in specialized environments. Raw human motion data, he notes, is like teaching a new language using material written for another one. Config’s answer is a dedicated data layer that preprocesses and transforms robot interaction data before it reaches training. This AI infrastructure startup claims more than 100,000 hours of human motion data already—over 30 times larger than leading public datasets—giving it a crucial head start in building the dense, task-rich corpora that general-purpose robot brains will rely on.
OEMs Race to Build Proprietary Robot Intelligence
The caliber of investors in Config’s seed round underlines how original equipment manufacturers now see robot AI as a core strategic asset. With backing from Samsung, Hyundai, LG, and others, Config is positioning itself as a neutral supplier of robot training data, analogous to how TSMC serves as a neutral foundry in semiconductors. Instead of licensing generic third-party AI, manufacturers can use Config’s robot AI data platform to develop proprietary robotic foundation models tuned to their own workflows, safety constraints, and hardware. Current customers span defense, agriculture, and industrial sectors, highlighting how physical AI is moving from pilots into real operations. As robots proliferate on factory floors and in fields, OEMs are preparing by securing access to unique data pipelines and infrastructure that can differentiate their automation offerings for decades.
Data Platforms Emerge as the Competitive Moat in Robotics
Config’s strategy reflects a broader industry realization: in robotics, high-quality data platforms are becoming the primary competitive moat. The company operates large-scale data production centers and plans to massively expand operations in Hanoi and Seoul, targeting 1 million hours of robot-relevant data. A key differentiator is its emphasis on converting and structuring data—rather than just training bigger models—so that robots can better perceive, predict, and act in complex environments. On top of this, Config is building an enterprise platform business and a cloud-based Robot-as-a-Service offering, allowing customers to tap its foundational models without expensive onboard compute. For AI infrastructure startups in robotics, this end-to-end stack—from raw data capture to processed datasets to cloud-delivered models—creates lock-in that is difficult for new entrants to replicate, and increasingly attractive to large industrial buyers.
What Config’s Funding Means for the Next Wave of Industrial Robots
With fresh capital and heavyweight OEM backing, Config is well placed to influence how the next generation of industrial robots is trained, deployed, and upgraded. Its ambition to be a neutral, large-scale supplier of robot-ready data gives manufacturers and system integrators an alternative to relying on closed, third-party models. As Config expands its dataset to cover more tasks and environments, its robotic foundation models could become a baseline layer that customers then customize for proprietary use. This could lower barriers to entry for traditional industrial businesses that want to integrate physical AI into existing supply chains without standing up their own data operations. If successful, Config’s approach will accelerate robot deployments across sectors while quietly shifting the center of gravity in robotics from hardware innovation to data and AI infrastructure.