[Edaily Reporter KIM JI-WAN ] IL SCIENCE CO.,LTD.(307180)is set to establish a production and demonstration hub encompassing its humanoid and physical AI businesses by leveraging this large-scale manufacturing infrastructure.

IL announced that it plans to build a smart factory using the manufacturing site of IL Mobility, a company it is in the process of acquiring as a subsidiary. The site covers approximately 7,341 pyeong (24,269 square meters), with a total floor area of approximately 3,555 pyeong (11,752 square meters). The building is planned to have one basement level and three above-ground floors. The company also released a rendering of the production facility designed based on this site.

This project is not merely focused on expanding production facilities. The vision is to combine the company’s existing automotive parts production capabilities with humanoid manufacturing, on-site application, operational data collection, AI training, and performance optimization to utilize the site as the core infrastructure for its Physical AI business.

In particular, the project is expected to gain momentum in tandem with IL Mobility’s transition to a wholly-owned subsidiary of IL. IL previously signed an agreement to acquire the remaining 610,000 shares of IL Mobility for 10 billion won. The scheduled date for the share acquisition is September 30. Once the transaction is finalized, IL Mobility will become a wholly-owned subsidiary of IL.

In a related regulatory filing, IL explained that the purpose of this share acquisition is to secure a demonstration base for the humanoid industry and to strengthen its competitiveness in physical AI operational data. This means that IL plans to utilize IL Mobility—which possesses experience in automotive parts production and factory infrastructure—as a practical test bed for its humanoid business.

The business model IL is envisioning differs from a simple distribution structure that involves importing and selling overseas humanoid products. The goal is to secure technology and products through collaboration with global humanoid companies while simultaneously building in-house capabilities for production, modification, verification, and operation tailored to the domestic manufacturing environment.

The key is a virtuous cycle in which data generated on the manufacturing floor is reused to improve robot performance. Deploying humanoids on the production floor generates a wide range of data during operations, including movement paths, performed actions, equipment status, and error or malfunction situations. IL plans to accumulate this data for AI training and then apply the results of that training to robot control and performance improvements.

In simplified terms, the business structure follows a cyclical process: “humanoid production → deployment to manufacturing sites → collection of operational data → AI training → performance enhancement → redeployment to the field.”

IL intends to implement this as a “Physical AI Data Flywheel.” The goal is to create a structure where performance continuously improves by integrating the robot hardware, the actual work environment, and the data generated on-site into a single AI system.

Another key objective of this project is to ensure that humanoids do not remain merely as robots for exhibitions or demonstrations, but actually perform tasks on real manufacturing lines. The company believes that by continuously training the AI with data collected during on-site deployment, work accuracy and efficiency will improve over time.

An IL Mobility official stated, “Our goal is to rapidly secure world-class technology through collaboration with global humanoid companies, while simultaneously building our own capabilities to produce and operate humanoids suited to domestic manufacturing sites,” adding, “We will establish a structure where data accumulated on-site drives AI learning and performance improvements to secure competitiveness in physical AI.”

He continued, “By integrating IL Mobility as a wholly-owned subsidiary and leveraging manufacturing infrastructure spanning approximately 7,300 pyeong, we plan to connect humanoids, manufacturing sites, operational data, and AI into a single business framework.” He added, “In the future, we will expand the scope of application to external manufacturing sites and broaden our business model to include RaaS (Robot as a Service) and other ventures that generate recurring revenue.”