NVIDIA Deepens Physical AI Partnerships Across South Korea With Samsung, LG, and Kakao Mobility
South Korean industrial giants are collaborating with NVIDIA to advance physical AI, deploying accelerated computing and robotics models in semiconductor fabrication, Level 4 autonomous driving, and factory automation.
South Korea's leading technology and manufacturing companies are accelerating the adoption of physical AI by deploying NVIDIA's computing architectures, simulation tools, and robotics foundations. Collaborations spanning Samsung Electronics, Kakao Mobility, and LG Electronics aim to move embodied AI and simulation-driven workflows from research laboratories directly into commercial production and real-world deployment.
Samsung Applies AI Physics to Chip Design and Materials
Samsung Electronics is expanding its engineering collaboration with NVIDIA to integrate accelerated computing and physics-informed AI into semiconductor research, process development, and device fabrication.
In technology computer-aided design (TCAD), Samsung uses NVIDIA cuISS and cuDSS libraries to speed up sparse linear system computations, alongside GPU acceleration for optical and process simulations. For materials discovery, Samsung’s sMOL platform recorded up to a 5x end-to-end speedup on selected quantum chemical calculations using NVIDIA cuEST, while evaluating NVIDIA cuEquivariance to accelerate tensor operations for atomistic AI models.
Additionally, Samsung is building chip-scale mechanical and thermal analysis pipelines with NVIDIA PhysicsNeMo. In an AI homogenization test case, combining AI-predicted material properties, coarser computational grids, and GPU acceleration yielded an approximate 1,600x speedup—slashing simulation time from about 18 hours on a 32-core CPU to 41 seconds on a GPU-accelerated coarse-grid run. Samsung also plans to evaluate the NVIDIA Agent Toolkit to orchestrate simulation models and support automated engineering design agents.
Kakao Mobility Targets Level 4 Autonomy With Kia Fleet
Kakao Mobility is teaming up with NVIDIA to build a Level 4 autonomous driving platform, pairing its proprietary autonomous software and operational data from Seoul’s Gangnam district with NVIDIA's automotive hardware stack.
The mobility company plans to deploy NVIDIA Hyperion, a sensor and compute reference architecture powered by the NVIDIA DRIVE AGX Thor system-on-a-chip. To ensure verification and safety across the development cycle, Kakao Mobility will integrate the NVIDIA Halos safety framework. Initial development work is targeted at Kia purpose-built vehicles (PBVs), setting the foundation for commercial Level 4 driverless mobility services across South Korea.
LG Demonstrates Factory Robotics Powered by Isaac GR00T
In industrial robotics, LG Electronics completed a demonstration of NVIDIA Isaac GR00T-powered manipulation tasks using its CLOiD robot directly on an active manufacturing floor.
The deployment established a closed-loop "physical AI data flywheel." The system collects real-world operational data, uses NVIDIA Cosmos for synthetic data augmentation, trains GR00T-based foundation models, and runs inference directly on the robot using onboard NVIDIA Jetson AGX Thor modules. LG and NVIDIA are using this simulation-to-real workflow to establish standardized processes for continuously training, testing, and refining industrial robots across manufacturing facilities.
