LG and NVIDIA Discuss Physical AI for Data Centers, Homes, Cars

LG and NVIDIA held talks in Seoul between LG CEO Ryu Jae-cheol and Madison Huang, senior director for Omniverse and Robotics at NVIDIA, to explore technical integration of NVIDIA software with LG hardware across data centers, homes and vehicles.

Participants focused on the hardware and compute requirements needed to run physical AI outside lab environments. Officials did not disclose investment amounts or timelines; discussions concentrated on technical alignment rather than commercial terms.

A central topic was rising compute density in data centers and its impact on cooling. NVIDIA’s GPUs have driven higher rack-level power use, and traditional air cooling can reach its limits as more compute is added per rack. LG showcased high-efficiency HVAC and liquid cooling systems for AI facilities at CES 2026, and the talks covered how those thermal systems could be paired with GPU infrastructure to prevent overheating and performance throttling.

In homes, the companies discussed local inference and simulation for consumer robots. LG has introduced CLOiD, a two-armed home robot with articulated hands that operates on an “Affectionate Intelligence” platform. Engineers described the need for near-zero-latency processing so cameras, local vector searches and control calculations can run on edge hardware to guide physical actions safely. NVIDIA’s Omniverse and Isaac robotics stack provide simulation environments and edge-compute software that can run inference locally and reduce ongoing cloud costs linked to continuous video ingestion and mapping.

NVIDIA has tested its robotics stack in industrial settings. A two-week Siemens trial announced at Hannover Messe featured a humanoid robot performing logistics tasks over an eight-hour period. Company representatives noted factory floors are more structured than private homes, and access to real domestic environments would yield variable lighting, clutter and human behavior data useful for training models. LG’s ThinQ ecosystem and wide consumer device distribution were discussed as sources of in-home data for model development.

Automotive topics included integrating LG’s in-cabin systems and infotainment with NVIDIA’s DRIVE compute platform. LG supplies EV components and in-cabin features such as gaze tracking and adaptive displays, while DRIVE provides vehicle autonomy compute. The companies discussed architectures that could standardize software layers, simplify API integration and support over-the-air machine-learning updates across fleets.

LG subsidiary LG CNS’s sponsorship of IoT Tech Expo North America was noted as part of LG’s enterprise activity. The discussions in Seoul are exploratory and focused on technical requirements for deploying physical AI; no formal agreements or schedules have been announced.