South Korea Unveils $1B Physical AI Roadmap to Modernize Factories

South Korea's Ministry of Science and ICT held a briefing in Seoul on September 16 to launch a physical AI research project covering North Jeolla and South Gyeongsang provinces, unveiling a five-year roadmap from 2026 to 2030 with a total investment of 1.4131 trillion won (about $1.01 billion). The initiative aims to extend AI from defect detection to production control, equipment coordination, and factory operations.

According to Seoul Economic Daily, the project marks the first time manufacturing companies' tacit knowledge and process experience have been incorporated into a national R&D program. The validated technologies will be expanded to manufacturing bases nationwide, with the goal of exporting a "K-Factory" model overseas.

The North Jeolla project, backed by 736.8 billion won, focuses on "factory orchestration" — enabling robots and equipment from different brands to work collaboratively and scheduling the entire plant as one system. The South Gyeongsang project, with 676.3 billion won, is developing a "large action model" that combines motion data from workers and robots, physical rules, and spatial information so AI can understand production processes and control real equipment.

In a proof of concept conducted in 2025, AI completed a robot factory design in about three hours, a task that previously took three to four experts nearly a month. So far, 13 companies have agreed to share tacit knowledge data. KAIST professor Chang Young-jae said Shinsung E&G has decided to invest 20 billion won in the physical AI ecosystem.

The government aims to combine accumulated manufacturing process data with AI, robotics, and digital twin technologies to deliver integrated solutions covering factory design, equipment coordination, and operations, potentially exporting entire AI-powered factory systems.