NVIDIA AI Day Seoul: Physical AI Enters Its Golden Age, With Samsung, Hyundai and LG on Board

What happened

The two-day AI Day Seoul 2026 opened on Nov 9 at COEX in Seoul's Gangnam district. Deepu Talla, NVIDIA's vice president of robotics and edge AI, said in the keynote that as agentic AI accelerates robot development, training and deployment, physical AI is entering a golden age and will become a key direction for the next phase of the AI industry.

Talla said robot development runs through three stages — training, simulation and real-world deployment — and NVIDIA supports the full path with DGX, RTX and Jetson computing platforms plus Omniverse digital-twin technology.

The Korean partnerships cluster into three lines: Samsung Electronics and SK Hynix are building digital twins of semiconductor fabs with Omniverse; Hyundai Motor Group is using Omniverse and Isaac for robot simulation in auto manufacturing; and LG Electronics is training its CLOiD robot on the GR00T platform for grasping and handling tasks.

Key facts

1. Samsung Electronics: Uses NVIDIA's PhysicsNeMo physics-AI tool for fluid, thermal and structural analysis to shorten chip design cycles
2. SK Hynix: Combines the CUDA-X accelerated computing library with PhysicsNeMo to shorten device and process simulation in memory manufacturing; it has been NVIDIA's principal memory partner since June 2026
3. Hyundai Motor Group: Is jointly developing robot simulation for auto manufacturing processes with NVIDIA, at speeds tens of thousands of times faster than real time
4. LG Electronics: Trains its CLOiD robot on the GR00T platform for object picking and handling
5. Kakao Mobility: Signed an MOU with NVIDIA on autonomous-driving technology to advance a Level 4 system in Korea, planning to integrate the Hyperion vehicle sensor and computing platform and Halos safety validation, starting with the Kia PBV dedicated vehicle
6. Rebellions and SK Telecom: Rebellions' R200 inference chip will connect to NVIDIA platforms through NVLink Fusion in MGX-design racks SK Telecom plans to run in its AI data centre, with systems deployable by 2028; NVIDIA GPUs handle prefill and the R200 handles decode
7. SK Telecom: Has committed to building a gigawatt-scale AI cloud on NVIDIA's DSX framework, with the first AI factory scheduled to come online in 2027