LG Group is building an AI factory that spans robotics, autonomous driving, data center cooling, and sovereign AI—and NVIDIA is providing the entire end-to-end stack.
That factory isn't a single building. It's a unified workflow: synthetic data generation using NVIDIA Cosmos, robot training in Isaac Sim, simulation and validation with Isaac Lab, and deployment on edge hardware. LG's existing manufacturing know-how from global sites gets plugged into a digital twin that connects raw material procurement through to customer delivery.
Home Robots Get a Digital Twin Workout
LG Electronics is developing CLoiD, a home cobot for indoor tasks. Instead of training on physical prototypes, LG will simulate the robot in physically accurate virtual environments using NVIDIA Isaac Sim and Isaac Lab. They're also evaluating the Isaac GR00T open reasoning vision-action language model to give CLoiD humanlike reasoning for complex tasks.
LG and NVIDIA plan to co-develop reference robots, putting LG's hardware into the Isaac GR00T ecosystem. To feed the training data pipeline, LG Electronics is building a physical AI data factory that uses Cosmos world foundation models for synthetic data generation and augmentation. LG Innotek will contribute sensing components optimized for NVIDIA's GPU architecture.
AI Factory Infrastructure Gets a DSX-Aligned Architecture
Beyond robotics, the AI factory infrastructure itself is getting an overhaul. LG Electronics is collaborating on cooling solutions—cooling distribution units (CDUs) and cold plates—for next-generation liquid-cooled AI factories. These align with NVIDIA's DSX AI factory platform, enabling rapid deployment of scalable supercomputing.
LG Uplus, the telecom arm, plans to build a large-scale AI data center capable of housing the latest NVIDIA GPUs. LG Energy Solution is working on 800-volt-direct-current energy solutions following NVIDIA's BESS Self-Qualification guidelines. LG CNS will integrate Isaac, Cosmos, and GR00T into its PhysicalWorks industrial robot platform for logistics and manufacturing.
Autonomous Driving and Sovereign AI on Blackwell
LG Electronics is aligning its ADAS and in-vehicle AI systems with the NVIDIA DRIVE platform, specifically DRIVE Hyperion for sensor, compute, and software architecture. DRIVE AGX compute will power AI cockpits and edge AI processing. LG Innotek plans to deliver next-gen components engineered for NVIDIA's architecture.
On the sovereign AI front, LG AI Research already used NVIDIA Blackwell GPUs, NeMo framework, and Nemotron open datasets to develop EXAONE, one of Korea's leading open LLMs. TensorRT-LLM optimized inference. Now LG Group is expanding EXAONE across its businesses via ChatEXAONE enterprise chatbot, with NVIDIA powering the underlying training and inference infrastructure.
What this means: LG isn't just buying NVIDIA GPUs. It's adopting the entire NVIDIA platform—from Cosmos for synthetic data to DSX for factory-scale deployment—as the backbone for every AI initiative across its family of companies. That's the kind of vertical integration that makes NVIDIA's ecosystem sticky, and it's exactly how a chaebol like LG will turn AI from a lab project into a factory-floor reality.
Source: NVIDIA and LG Group Build an AI Factory to Advance Physical AI, Mobility and AI Infrastructure
Domain: blogs.nvidia.com
Comments load interactively on the live page.