NC AI will take part in a 1.4 trillion won ($1 billion) government research and development program on physical artificial intelligence in North Jeolla and South Gyeongsang provinces. The company will develop technology that links AI training in virtual environments to autonomous control of equipment and robots on factory floors, drawing on digital twins, world models and physical AI foundation models.
NC AI said on the 17th that it had been selected for tasks in both regions under the physical AI R&D program run by the Ministry of Science and ICT and the National IT Industry Promotion Agency. The state project will invest a total of 1.4131 trillion won over five years from 2026 to 2030, developing autonomous factory technology in North Jeolla and precision control of manufacturing processes in South Gyeongsang.
In North Jeolla, NC AI will handle two tasks. Through a project on virtual environment technology for collaborative-intelligence physical AI digital twins, it will recreate factory equipment, work spaces and worker movements in three-dimensional virtual environments. The company plans to generate AI training data in that space and enable the training and validation of vision-language models and vision-language-action models.
In a separate task on foundation models specialized for unmanned factory operation and physical AI-based software-defined factories, NC AI will develop a physical AI foundation model that recognizes spaces and situations on manufacturing sites. Using large language models and vision-language models, it will build a base that identifies static and dynamic conditions inside a plant and allows equipment and robots to respond autonomously.
In South Gyeongsang, the company will take charge of developing physical intelligence action models and an integrated framework. Based on field data from sensors and vision systems, it will develop a world model that identifies and predicts physical conditions and changes, and use large action models to extend that into action planning and control of equipment and robots. The technology will be applied to manufacturing areas such as shipbuilding welding and semiconductor back-end processes.
NC AI has secured physical AI technologies ranging from training and validation in virtual environments using digital twins, to understanding physical conditions through world models, to controlling equipment and robots with action models and robot foundation models. This year, the company has pursued related technology development and demonstrations with Hanwha Ocean, POSCO DX and Hyundai Rotem. In shipbuilding, it is working on autonomous welding based on collaborative robots; in defense, on virtual battlefield environments and robot training based on world models; and in steel, on behavior training for humanoid robots.
"Strengthening manufacturing competitiveness through physical AI is an important task for the future of domestic industry," NC AI CEO Lee Yeon-soo said. "We will apply AI trained in virtual environments to actual manufacturing sites and validate the technology in various industrial settings, laying a foundation that can lead the AI transformation of Korea's manufacturing sector."
