LG, NVIDIA Explore Data Center Cooling, Robotics and In-Car AI

In Seoul, LG CEO Ryu Jae-cheol met with Madison Huang, senior director for Omniverse and Robotics at NVIDIA, to discuss physical AI applications and the hardware and software needed to run them.

The companies discussed data center cooling tied to higher power density in modern server racks. NVIDIA’s data center revenue has grown rapidly, and higher power density is creating thermal limits for conventional air cooling. When racks overheat, servers reduce performance. LG showed high-efficiency HVAC and thermal-management systems it displayed at CES 2026 and described integrating cooling hardware directly into data center designs to increase compute per square foot while maintaining safe operating temperatures.

LG is preparing commercial units to supply thermal hardware into AI infrastructure. LG CNS, the company’s IT arm, is listed as a sponsor of IoT Tech Expo North America.

The talks also covered edge robotics and low-latency inference. LG recently introduced CLOiD, a home robot with two seven-degree-of-freedom arms and multi-fingered hands running on an “Affectionate Intelligence” platform for contextual learning. Home manipulation requires near-zero latency: a robot must process live camera input, query local databases for object properties and compute grip forces in real time. LG does not yet have a full set of digital twins, pre-trained manipulation models and simulation tools to accelerate safe home deployment.

NVIDIA supplies Omniverse and the Isaac robotics stack for real-time inference and simulation. Moving more processing to local devices would reduce cloud costs and bandwidth from continuous video and spatial mapping, and could shorten the path from prototype to commercial product using simulation-to-reality pipelines.

NVIDIA tested its robotics stack in industrial settings in January 2026 in a two-week Siemens factory trial that ended with a humanoid robot performing live logistics tasks over an eight-hour run; the result was announced at Hannover Messe in April. Company representatives noted factory floors are more structured than private homes, which vary in lighting, clutter and human behavior. LG’s ThinQ ecosystem and retail channels could supply diverse real-world data for training models for domestic environments.

Automotive integration was another topic. LG’s automotive components unit supplies in-vehicle infotainment, EV components and in-cabin features such as gaze tracking and adaptive displays. NVIDIA’s DRIVE platform is used for vehicle autonomy and advanced driver-assistance computing. The companies discussed linking LG’s interior experience layer with NVIDIA’s vehicle compute stack to offer a unified reference architecture for automakers and to enable over-the-air machine-learning updates to production fleets.

The firms have not announced formal investments, binding agreements or specific project timelines. Discussions remain exploratory and focused on aligning hardware and software requirements for physical AI deployment.