AI Revolution 🚀🔥: Asia's New Robot Power!

THE RISE OF PHYSICAL AI MANUFACTURING
The burgeoning artificial intelligence landscape in Asia, particularly within nations like South Korea, Japan, China, and Taiwan, is intrinsically linked to the region’s established manufacturing dominance. These economies, traditionally reliant on large-scale production, export-driven industries, and highly optimized supply chains, are now fueling a significant shift in AI investment and adoption. This structural advantage provides a unique foundation for the development and deployment of AI, especially in the rapidly growing field of robotics.

CONFIG: A DATA LAYER POWERHOUSE
Config, a Seoul- and San Jose-based startup, is at the forefront of this movement, specializing in the data layer for robotic foundation models (RFMs). The company’s recent $27 million seed round, oversubscribed by major manufacturers like Samsung, Hyundai, and LG, demonstrates the substantial interest in their approach. Strategic investors including SKT America and various Korean venture capital firms, alongside prominent figures like Pieter Abbeel, further validate Config’s potential. Founded in January 2025 by Minjoon Seo and a team with backgrounds in Meta, Twelve Labs, Waymo, Google, and Naver, Config’s core mission is to provide the data robots need to learn and operate effectively.

THE UNIQUE CHALLENGES OF ROBOTIC AI TRAINING
Unlike the relatively straightforward process of training large language models – which relies on readily available, vast amounts of text data – training robots presents a significantly more complex and costly hurdle. Teaching robots to move requires physically collecting training data, encompassing the robot itself, the operational environment, and human operators. Minjoon Seo highlighted this disparity, stating that robotics AI development is currently more expensive than software-only chatbot development due to the intensive data collection process. This emphasis on physical data acquisition contributes to the escalating costs as companies pursue increasingly capable robots.

DATA AS THE NEW CHIP: CONFIG’S STRATEGIC ROLE
Config’s business model mirrors that of TSMC, a leading chip manufacturer that produces for major tech giants like Apple, Nvidia, and AMD without competing directly with them. Similarly, Config aims to be the premier supplier of robotics AI data, effectively acting as a critical infrastructure provider. The company’s strategy is gaining traction as manufacturers increasingly seek to develop proprietary robot AI systems independently, rather than relying solely on external vendors. This represents a substantial market opportunity that Config is strategically positioned to capitalize on.

REVENUE GENERATION AND KEY CUSTOMERS
Despite being a relatively young startup, Config is already generating revenue, according to COO and co-founder Jack Bang. Their current customer base includes large manufacturers, system integrators, and companies operating in the agriculture and defense sectors. This diversified customer base demonstrates the broad applicability of their data solutions. Config’s approach is further distinguished by its significant data accumulation – boasting over 100,000 hours of human motion data, surpassing the approximately 3,000 hours available in the largest comparable open-source dataset, AgiBot World.

TRANSFORMING DATA BEFORE TRAINING
Config’s innovative approach involves transforming the data before training begins, tailoring it specifically to the way robots move and interact with the world. This mirrors the process of language translation, where data is converted rather than the model itself. Seo emphasizes this as Config’s core technical differentiator, a crucial step in ensuring the seamless integration of data with robotic AI systems.

SCALING DATA PRODUCTION: A GLOBAL FOOTPRINT
The startup operates out of Seoul and Hanoi, supporting a workforce of nearly 300 dedicated to data production. This global footprint allows Config to efficiently capture and process data from diverse environments and applications, further bolstering the quality and scope of their offerings. This robust infrastructure underpins their ambition to become the dominant provider of robotics AI data.