Config Intelligence Inc.
Category: Robotics / Embodied AI
Building the data infrastructure and foundation models for general-purpose bimanual robotics Config Intelligence Inc. was founded in 2025. The company is led by Minjoon Seo. Based in Seoul, South Korea & San Jose, California, USA. Team size: 201-500. Total funding raised: $35M. Latest round: Seed. Key investors include Samsung Venture Investment; Hyundai Motor ZER01NE Ventures; LG Technology Ventures; SKT America; Mirae Asset Ventures; Korea Development Bank; GS Futures; Kakao Ventures; Z Ventures; Pieter Abbeel (angel).
- Founded
- 2025
- Headquarters
- Seoul, South Korea & San Jose, California, USA
- Team size
- 201-500
- Total funding
- $35M
Value proposition
Config provides the data layer for robotic foundation models (RFMs) — collecting, transforming, and supplying high-quality human motion data so robotics companies can train capable bimanual robots without building their own data infrastructure. Positioned as the "TSMC of robot data."
Products and solutions
Config Data Platform (end-to-end pipeline from raw data to model evaluation); CFG-1 (flagship 2B-parameter foundation model for bimanual manipulation with native 3-minute memory, <50ms latency on RTX 5090); Robot-as-a-Service (cloud-based foundation model offering); 100,000+ hours of human motion data
Unique value
Config transforms human motion data into robot-ready training data before model training begins (data conversion technology, not model adaptation) — analogous to translating language rather than expecting the model to learn cross-lingually. Operates as a neutral data supplier (TSMC model) without competing with its customers' robot hardware businesses.
Target customer
Large manufacturers, system integrators, agriculture companies, defense sector organizations building proprietary robot AI
Industries served
Manufacturing, Agriculture, Defense, System Integration
Technology advantage
Proprietary data conversion technology that transforms human motion data to be inherently compatible with robot kinematics before training; 100,000+ hours of in-house collected human motion data (30x larger than AgiBot World, the largest comparable open-source dataset); CFG-1 foundation model with 2B parameters optimized for efficient edge inference; long-horizon memory architecture supporting 3-minute task sequences; micrometer-level action estimation precision
How they differentiate
Unlike competitors who build full-stack robot hardware + AI (e.g., Physical Intelligence, Skild AI), Config operates purely as a data infrastructure and foundation model provider — the "TSMC of robot data" — supplying data and models to any robotics company without competing with them. Config's core differentiator is its data conversion technology that transforms human motion data to robot-compatible formats before training, rather than adapting models post-training.
Main competitors
Physical Intelligence; Skild AI; Generalist AI
Key partnerships
Samsung Venture Investment (lead investor); Hyundai Motor ZER01NE Ventures; LG Technology Ventures; SKT America; Mirae Asset Ventures; Korea Development Bank; GS Futures; Kakao Ventures; Z Ventures
Notable customers
Large manufacturers, system integrators, agriculture companies, and defense sector organizations (specific names not publicly disclosed)
Major milestones
Founded January 2025; Raised $8M pre-seed in 2025; Launched Tech Preview (Feb 2026); Raised $27M seed at $200M+ valuation led by Samsung Venture Investment (May 2026); Accumulated 100,000+ hours of human motion data; Launched CFG-1 foundation model; Announced Robot-as-a-Service product
Growth metrics
Already generating revenue; targeting $10M ARR by end of 2027; 100,000+ hours of human motion data collected; ~300 employees in data production (Seoul + Hanoi); valued at $200M+ post-money
Market positioning
Positioned as the neutral data layer provider for the robotics AI ecosystem, analogous to TSMC in semiconductors. Backed by Korea's largest manufacturing conglomerates (Samsung, Hyundai, LG, SK), giving it unique access to industrial robotics deployment environments and strategic manufacturing partnerships in Asia.
Geographic focus
Asia-Pacific (South Korea, Vietnam for data operations), with US presence in San Jose, CA. Targeting global robotics companies building proprietary robot AI.
About Minjoon Seo
Ex-Meta Researcher; Chief Scientist at TwelveLabs; Associate Professor at KAIST AI. PhD from University of California, Berkeley.
Official website: https://config.inc