NVIDIA and LG Partner to Build Massive Physical AI Factories
The next robot to load your dishwasher or vacuum your hallway will likely be trained in a virtual world long before it reaches your home. That is the idea behind a new partnership in which NVIDIA and LG Group are building an AI factory together, turning LG's decades of global manufacturing data into working physical AI. According to NVIDIA, the NVIDIA LG AI factory will give LG the accelerated computing infrastructure to train, simulate, validate and deploy AI applications across robotics, autonomous driving, data center technology and GPU cloud services.
The setup connects several pieces that usually live in separate teams. NVIDIA says the collaboration links AI model development, physical AI data generation, robot simulation, edge deployment and factory-scale digital twins into one workflow. For everyday users, that pipeline is what decides whether a home robot actually works safely in a messy, unpredictable living room rather than a tidy lab.
From Home Robots to Synthetic Training Data
LG Electronics is developing home robots like CLOiD for everyday indoor tasks, with the stated goal of improving convenience and quality of life. By folding NVIDIA Isaac Sim and Isaac Lab into its workflow, the company can train and validate those cobots in physically accurate virtual environments before a single unit reaches a family's home. LG is also exploring the NVIDIA Isaac GR00T reasoning model to give its robots humanlike decision-making for complex tasks.
Training data is the hard part of robotics, and it directly shapes how well these machines behave around people. According to NVIDIA, LG Electronics is building a physical AI data factory that uses NVIDIA Cosmos world foundation models to generate and augment synthetic data. The goal is to supply high-quality training data to Korean and global companies, not just LG's own projects.
Other LG units are pulling in the same direction. LG Innotek plans to supply sensing components tuned for NVIDIA's GPU architecture, while LG CNS is wiring Isaac frameworks, Cosmos models and GR00T into its PhysicalWorks platform to bring AI robots onto manufacturing and logistics floors, where they take on repetitive and heavy work alongside people.
Mobility, Infrastructure and Sovereign AI
On the road, LG Electronics is aligning its driver-assistance and in-vehicle systems with the NVIDIA DRIVE platform. The work focuses on matching sensor, compute and software design to the NVIDIA DRIVE Hyperion architecture, with LG also planning to use NVIDIA DRIVE AGX compute for AI-powered cockpits and edge processing in future vehicles. For drivers, that points toward cars that assist more and demand less.
The infrastructure side is just as concrete. LG Electronics is contributing cooling distribution units, cold plates and prefabricated modular designs aligned with the NVIDIA DSX AI factory platform. LG CNS and LG Uplus plan to build scalable, power-efficient AI factories on NVIDIA DSX, and LG Uplus is also building a large-scale AI data center to house the latest NVIDIA GPUs. LG Energy Solution is working with NVIDIA on 800-volt direct-current energy solutions for next-generation GPUs.
NVIDIA and LG AI Research are also advancing EXAONE sovereign AI, one of Korea's leading open model families. According to NVIDIA, LG AI Research used NVIDIA Blackwell GPUs, the NeMo framework and Nemotron datasets to develop EXAONE, plus TensorRT-LLM for optimized inference. LG is now exploring wider use through ChatEXAONE, its enterprise chatbot service, which puts the technology in the hands of everyday office workers.
What to watch next is execution. The companies have outlined an ambitious stack across robots, vehicles, data centers and sovereign models, and the real test will be whether it shows up in homes, cars and workplaces people actually use. You can read the full announcement on the NVIDIA blog.