NVIDIA CEO Jensen Huang landed in Seoul on June 6 for what the Korea Herald called his longest visit to South Korea in recent years. The trip was explicitly about aligning the AI supply chain, and it produced a steady stream of partnership announcements.
The two biggest ones: LG Group and Doosan Group each signed deals to build what NVIDIA calls AI factories. These are not traditional production facilities. They are computing centers packed with NVIDIA processors, purpose-built to train robots, simulate real-world environments, and deploy AI systems across industrial operations.
LG's scope is wide. The group is building home robots, training them inside virtual environments before releasing them into real homes. It is developing logistics and industrial robots through its LG CNS business unit. Its automotive components division is aligning sensors and software with NVIDIA's self-driving car platform. LG AI Research is also developing EXAONE, a Korean-language AI model, using NVIDIA hardware. And LG Uplus, its telecom arm, is planning a large-scale data center built around NVIDIA's latest processors.
Doosan's angle is heavier industry. Doosan Robotics is building what it calls an Agentic Robot OS, a software layer that lets industrial robots perceive, reason, and act across complex tasks like sorting warehouse pallets or sanding manufactured parts. Doosan Bobcat, known for compact construction equipment, is exploring how to make its machines more autonomous for construction, agriculture, and landscaping. Doosan Enerbility, which builds gas turbines and is exploring small nuclear reactors, is looking at powering NVIDIA's own data centers.
These two deals sit inside a larger picture. Back in October 2025, during the APEC summit in Gyeongju, NVIDIA announced commitments across South Korea totalling over 260,000 processors. Samsung is building a facility with more than 50,000 units for semiconductor manufacturing. SK Group is building one of similar scale. Hyundai Motor Group committed roughly 3 billion dollars jointly with NVIDIA and the Korean government to advance autonomous vehicles and smart factories.
What is actually happening here is that NVIDIA is turning South Korea into its largest outside testing ground for physical AI. Korea is a good fit: it has world-class manufacturing expertise, strong industrial conglomerates that span many sectors, and a government actively co-investing in the build-out.
For non-Korean businesses watching this, the practical implication is not abstract. Physical AI, which means machines that can observe and react to the real world rather than just follow fixed rules, is moving from pilot projects into commercial rollout. One analyst forecast puts the market at roughly 5 billion dollars today, growing to nearly 50 billion by 2033. The Doosan and LG deals suggest the rollout is accelerating, not slowing.
The most direct effect on global businesses will come through supply chains and competition. If a company buys components from Korean manufacturers, those suppliers are now building towards more automated, AI-driven production. Lead times, quality consistency, and cost structures will shift. If a company competes with Korean manufacturers in any industrial category, the gap between early adopters and everyone else in automation is widening.
The Korea deals also tell you something about NVIDIA's strategy. The company is not waiting for industries to come to it. It is embedding itself directly into the infrastructure, the training data pipelines, the robot operating systems, and the power solutions of the world's largest manufacturers. Every layer of the deal with LG and Doosan touches a different part of what it takes to build and run intelligent machines at scale.
That kind of depth is hard to reverse once it is in place.