LG Electronics (LG), a South Korean multinational conglomerate, and NVIDIA are expanding their AI partnership by establishing an AI factory that will provide LG Group with computing and data infrastructure for training, simulation, validation and deployment of AI applications. Earlier this month, the companies signed a memorandum of understanding (MOU) covering the broader partnership. This was followed by a visit by NVIDIA officials to LG’s Data Factory, as the companies continued work on the AI infrastructure.
The LG Data Factory is under construction at the company’s Yangjae R&D Campus in Seoul, and the facility will have four floors covering 10,000 square meters to house several hundred robots. It is designed to collect and process data generated as robots perform tasks in simulated homes and manufacturing environments.
According to a press release, the factory is intended to support AI development across LG’s businesses, including robotics, manufacturing, autonomous driving and data centers. The work combines NVIDIA’s accelerated computing and software platforms with LG’s hardware, manufacturing and operational data.
LG Electronics is developing data center infrastructure for the AI factory, including cooling distribution units, cold plates and prefabricated modular designs aligned with NVIDIA’s DSX AI factory platform. LG CNS plans to develop AI factories using NVIDIA GPUs, while LG Uplus is planning a large-scale AI data center capable of supporting NVIDIA’s latest GPUs and future GPU cloud services. LG Energy Solution is working with NVIDIA on 800-volt direct-current power systems for data centers.
Lyu Jae-cheol, CEO, LG Electronics, said, “Through the synergy built on ‘One LG’ bringing together core capabilities across the Group and strategic collaboration with global partners, we will secure our competitiveness in physical AI and become a comprehensive robotics solutions provider.”
LG expects the facility to be fully operational by the end of 2026, and generate 100,000 hours of training data. The figure includes data collected directly from robots as well as synthetic data generated and augmented using NVIDIA Cosmos which LG estimates the total is equivalent to 12 years of data.
The facility includes a simulated home where LG Electronics’ CLOiD robots can learn and repeat cleaning tasks, and a manufacturing environment based on LG’s washing machine plant in Tennessee. LG is also collecting data from logistics automation systems operated by LG CNS and from LG Innotek’s robotic-hand training environment.
The resulting datasets will be used to develop LG’s Robot Foundation Model, which is intended to provide capabilities for humanoid robots. NVIDIA technologies being used in the data-generation and development process include Omniverse libraries, Cosmos and the Isaac robotics platform.
LG CNS is integrating NVIDIA Isaac, Cosmos and GR00T technologies into its PhysicalWorks platform for manufacturing and logistics. LG and NVIDIA are also working on digital twins that can combine factory data with simulated environments to test and optimize production and logistics operations.
The infrastructure will also support automotive AI, LG Electronics plans to use NVIDIA DRIVE Hyperion and DRIVE AGX for advanced driver-assistance systems, AI-powered cockpits and edge AI applications. LG Innotek is developing sensing, connectivity and lighting components for autonomous driving systems based on NVIDIA architecture. LG AI Research is separately using NVIDIA Blackwell GPUs, the NeMo framework, Nemotron datasets and TensorRT-LLM to develop its EXAONE sovereign AI models.