[One Year After the Kkanbu Meeting (2)] Turn Broader AI Cooperation into Homegrown Technology

Hyundai Motor and LG Turn Shop-Floor Experience into Training Data for Factories and Robots
Samsung Electro-Mechanics, LG Innotek and Others Join NVIDIA's Supply Chain
Building Up Technology and Strengthening Manufacturing Competitiveness Remain Key

Since last year's "Kkanbu meeting," cooperation between Korean companies and NVIDIA in artificial intelligence (AI) has expanded from memory to factories and robots. Korean companies are applying NVIDIA's technology on production floors and using data and component technologies accumulated through manufacturing to develop robots. If securing graphics processing units (GPUs) was the starting point of the alliance, strengthening the competitiveness of Korean manufacturing through AI has now emerged as a key priority.

◆From adopting GPUs to jointly developing factories and robots= LG Electronics is working with NVIDIA to use manufacturing sites to train robots. It collects real-world task data at its Data Factory in Yangjae, Seoul, and synthesizes virtual data using NVIDIA technology. Its goal is to amass 100,000 hours' worth of robot training data by the end of the year by combining real and virtual data.

At its washing machine plant in Tennessee, U.S., and its production line in Changwon, South Gyeongsang Province, the company is testing "LG CLOiD" to validate tasks such as transporting parts, assembly, and packaging. The effort aims to turn manufacturing experience accumulated in factories into robots' ability to perform tasks.

Hyundai Motor and Kia are also developing technology that uses NVIDIA's "Isaac Sim" to recreate real factories and robot movements in virtual spaces and verify equipment operations in advance. They are also developing technology that uses design drawings and simulations to map robots' work paths and cameras to correct discrepancies with their actual positions. This would reduce the time needed to deploy robots, a process that previously required people to enter each work point one by one. Development is targeted for completion in June 2027.

◆Beyond HBM, reaching substrates and the eyes of robots= The range of products Korean companies supply to NVIDIA is also expanding. It is extending beyond high-bandwidth memory (HBM), a key component of AI semiconductors, to semiconductor substrates and cameras for robots.

On October 21, Samsung Electro-Mechanics announced at GTC in Berlin, Germany, that it had joined NVIDIA's supply chain for substrates used in AI semiconductor packages. The two companies agreed to jointly invest in production facilities in Sejong and Busan and to establish a long-term supply arrangement for the substrates.

At the same event, LG Innotek also unveiled robot cameras and a multimodal vision system incorporating NVIDIA technology. It is expanding optical technologies developed for smartphones and automobiles into technologies that enable robots to recognize their surroundings and objects.

This is an example of how the competitiveness of Korean manufacturing is being translated into components and technologies needed by the AI industry. It points to the possibility that Korean companies' role, previously focused on supplying memory, could expand into robotics and manufacturing AI.

Hyundai Motor Group Executive Chair Euisun Chung, NVIDIA CEO Jensen Huang, and Samsung Electronics Chairman Lee Jae-yong smile brightly on stage at COEX in Seoul on October 30 last year. Photo by Paek Jongmin, Tech Specialist.

◆Turning the gains from cooperation into homegrown technology= The key is to build on cooperation with NVIDIA to develop Korean companies' own technologies and products. Beyond introducing AI in factories, companies must use on-site data and skilled workers' experience to increase productivity and improve quality.

"Polaris," a manufacturing-focused AI agent unveiled by Hyundai Motor and Kia from October 19 to 21, points in this direction. It connects production and quality information with past equipment failure cases and skilled workers' know-how to identify the causes of equipment abnormalities and suggest corrective actions. Lim Yongtaek, a senior manager on Hyundai Motor's Manufacturing AI Technology Development Team, said, "Polaris can analyze the causes of equipment abnormalities within minutes, whereas the process used to take a long time."

Efforts to broaden the foundation for cooperation have also begun. In July 2026, Hyundai Motor Group and NVIDIA announced plans to jointly build a "Robot Reference Platform" that will provide universities, research institutes, and startups with research robot models and a common development environment. Euisun Chung, Executive Chair of Hyundai Motor Group, stressed, "The results of cooperation with big tech companies must help lay the groundwork for the growth of Korea's physical AI industry."

This content was produced with the assistance of AI translation services.