Summary
NVIDIA describes collaborations with Samsung Electronics, Kakao Mobility and LG Electronics spanning chip engineering, autonomous-driving development and factory robotics. The work ranges from planned development to a robot demonstration in a manufacturing environment.
NVIDIA says collaborations with Samsung Electronics, Kakao Mobility and LG Electronics are applying accelerated computing, AI and simulation to chip engineering, autonomous-driving development and factory robotics. The projects described in the company’s November 8, 2026 announcement are at different stages: Kakao Mobility is planning development work, while NVIDIA and LG Electronics have demonstrated robotic tasks in a manufacturing environment.
“Physical AI” is a broad term for AI used in systems connected to the physical world, such as robots and vehicles, as well as the engineering simulations used to design them. The Korean projects illustrate different parts of that pipeline, from modelling materials and devices to training and evaluating robots.
Faster simulation for semiconductor engineering
Samsung Electronics is expanding work with NVIDIA on materials research, process development and device engineering. In technology computer-aided design, the companies are using NVIDIA cuISS and cuDSS to accelerate calculations involving sparse linear systems—large equation systems in which many coefficients are zero and which are common in engineering simulations. The collaboration also includes GPU acceleration for process and optical simulations.
NVIDIA says Samsung’s sMOL platform has run selected quantum-chemistry calculations up to five times faster end to end using NVIDIA cuEST. Samsung is also exploring NVIDIA cuEquivariance to accelerate tensor operations used in training and running atomistic AI models, which represent materials at the level of their atoms.
For chip-scale thermal and mechanical analysis, Samsung is developing workflows with NVIDIA PhysicsNeMo. NVIDIA reports that one AI homogenization case—where effective material properties are used to represent material behaviour at a coarser scale—took about 41 seconds on a GPU-accelerated coarse grid, compared with about 18 hours for a fine-grid solve on a 32-core CPU. The company describes this as a roughly 1,600-fold speedup. Because the comparison uses different hardware and grid resolutions, it is a result for that particular workflow rather than a direct comparison of identical calculations.
Robots and vehicles move toward real-world use
NVIDIA and LG Electronics demonstrated robotic manipulation and manufacturing tasks using LG’s CLOiD robot in an actual manufacturing environment. Their workflow links real-world data collection with data augmentation using NVIDIA Cosmos, training with NVIDIA Isaac GR00T, onboard computing with NVIDIA Jetson AGX Thor and simulation-based evaluation. The aim is to create a repeatable process for training, evaluating and improving the robot’s capabilities.
Kakao Mobility, meanwhile, plans to develop a Level 4 autonomous-driving system using its driving software and experience operating in Seoul’s Gangnam district alongside NVIDIA vehicle technologies. The planned setup includes NVIDIA Hyperion, a reference platform for vehicle sensors and computing, powered by NVIDIA DRIVE AGX Thor; Kakao Mobility also plans to use NVIDIA Halos for safety work during development and validation. Initial development is planned around Kia purpose-built vehicles. The stated goal is to support future Level 4 mobility services in Korea.
Taken together, the work shows how AI is being applied not only to generate predictions, but also to support engineering simulations and systems that sense or act in physical environments. The announcement describes a mix of research and development, planned vehicle-system work and a factory demonstration—not a single, uniformly deployed industrial platform.