THE CRUNCH
NVIDIA says it is working with Korean industry leaders to move physical AI, meaning AI systems that operate in the real world, from development into production. The collaborations span three very different domains: semiconductor engineering at Samsung, autonomous driving at Kakao Mobility and robot learning at LG Electronics.
Samsung is expanding its work with NVIDIA on chip development, using tools such as cuISS and cuDSS to speed up sparse linear system computations in technology computer-aided design, plus GPU acceleration for process and optical simulations. Its sMOL materials platform has run selected quantum chemical calculations up to 5x faster end to end with NVIDIA cuEST, and the company is exploring cuEquivariance for atomistic AI models.
The most striking figure comes from an AI homogenization case that NVIDIA and Samsung report: combining AI-predicted material properties, coarser grids and GPU acceleration cut simulation time from about 18 hours on a 32-core CPU to 41 seconds, a roughly 1,600x speedup. Samsung is also developing chip-scale thermal and mechanical analysis workflows with NVIDIA PhysicsNeMo, and NVIDIA's Agent Toolkit could eventually connect these tools into engineering agents.
Kakao Mobility plans to develop a level 4 autonomous driving system, one that handles all driving within its operating area, using NVIDIA's Hyperion reference platform powered by DRIVE AGX Thor and the NVIDIA Halos safety framework. Initial development is planned around Kia purpose-built vehicles, drawing on Kakao Mobility's experience operating in Seoul's Gangnam district.
LG Electronics demonstrated a CLOiD robot performing manipulation and manufacturing tasks in a real factory environment, powered by NVIDIA Isaac GR00T. The teams built a data flywheel linking real-world data collection, Cosmos-based augmentation, model training, Jetson AGX Thor onboard computing and simulation-based evaluation, intended as a repeatable process for improving manufacturing robots.