NVIDIA and LG Group outline an “AI factory” roadmap spanning robotics, mobility, sovereign AI and next-generation data centers

NVIDIA said it is working with LG Group to build an “AI factory” intended to speed the development and deployment of physical AI systems, from household robotics and factory automation to autonomous driving and AI cloud services. Announced June 7, 2026, the effort is framed as a unified workflow that connects AI model development, physical AI data generation, robot simulation and training, edge deployment, and factory-scale digital twins. The companies say the goal is to give LG accelerated computing infrastructure for training, simulating, validating, and deploying AI across its businesses.

LG’s stated advantage in the collaboration is its production technology data and know-how drawn from global manufacturing sites, which NVIDIA says can be combined with NVIDIA’s AI infrastructure and digital twin technologies. The companies describe plans to create an autonomous manufacturing ecosystem intended to connect key factory steps, from raw material procurement to production, logistics, and customer delivery, through real-time data and AI. They also say the target is to establish a “global smart factory standard.”

On the robotics side, LG Electronics is developing home-based robots such as CLoiD. NVIDIA says LG plans to integrate NVIDIA Isaac Sim and NVIDIA Isaac Lab, two open robotics simulation tools that NVIDIA describes as reference frameworks for robotics simulation and for robot learning workflows. The integration is aimed at simulating, training, and validating cobots in physically accurate virtual environments before deployment.

The partners also referenced NVIDIA Isaac GR00T, described in NVIDIA materials as an open “reasoning vision action language model” for robot skills and task execution. NVIDIA said LG is exploring GR00T for its home robots and modular robotics platforms, with plans to jointly develop reference robots positioned as part of the GR00T ecosystem. In parallel, LG Electronics said it is developing a “physical AI data factory” to reduce robotics training-data bottlenecks by using NVIDIA Cosmos world foundation models for synthetic data generation and augmentation.

Beyond simulation and synthetic data, the collaboration extends into components and operational adoption. NVIDIA said LG Innotek plans to supply robotics components, including sensing solutions, designed to align with NVIDIA’s development environments and GPU architecture. NVIDIA also said LG CNS is building an ecosystem to help manufacturing and logistics operators adopt AI robots, using LG CNS’s PhysicalWorks platform while integrating NVIDIA robotics technologies, Cosmos world models, and the Isaac GR00T foundation models into that system.

For the underlying compute and facilities, NVIDIA said the project is aligned with its DSX AI factory approach, described by NVIDIA as a co-designed system across chips, networking, power, cooling, and operations rather than a set of disconnected components. The company said LG Electronics will expand technical collaboration beyond cooling certification, including work on prefabricated modular design capabilities intended to address power, thermal, and deployment requirements for liquid-cooled AI factories. NVIDIA described DSX as coordinating compute, cooling, power and operations within a unified architecture.

The companies also tied the “AI factory” effort to power and energy systems. NVIDIA said LG Uplus, working with LG Electronics and LG Energy Solution, plans to build scalable, power-efficient AI factories based on NVIDIA DSX and to develop a large-scale AI data center capable of accommodating the latest NVIDIA GPUs. LG Energy Solution is also expected to collaborate on emerging 800-volt DC data center energy solutions, in alignment with NVIDIA’s BESS Self-Qualification guidelines, which NVIDIA characterizes as a partner-run process for qualifying battery energy storage systems for grid buffering and demand response use cases.

NVIDIA said the partnership includes a mobility component focused on integrating LG Electronics’ advanced driver-assistance systems and in-vehicle AI systems with NVIDIA’s DRIVE platform. NVIDIA described DRIVE Hyperion as a production-ready autonomous driving development platform and reference architecture that combines a standardized sensor suite with NVIDIA DRIVE compute and a software stack. NVIDIA added that LG Electronics plans to use NVIDIA DRIVE AGX accelerated compute for mobility applications including AI-powered cockpits and edge AI processing.

Finally, NVIDIA and LG AI Research said they are collaborating to advance EXAONE, described as a Korea-focused sovereign AI model and an open model family available for developers, enterprises and researchers. NVIDIA said LG AI Research used NVIDIA Blackwell GPUs, the NVIDIA NeMo framework, NVIDIA Nemotron datasets, and NVIDIA TensorRT-LLM software to support EXAONE development and to build high-performance inference engines for optimized deployment. The companies also pointed to ChatEXAONE, an EXAONE-based enterprise chatbot service, as a vehicle for wider adoption of EXAONE and “agentic AI” technologies across LG businesses.

What is not disclosed in the announcement is as significant as what is. NVIDIA does not provide timelines, site locations, capacity figures for data centers, target performance benchmarks, or any financial terms tied to the facilities, components, or cloud services it describes. It also does not specify which exact LG product lines will become first deployments of the robotics and digital-twin workflow, or how quickly open frameworks and reference models will transition into production operations. For LG and NVIDIA watchers, the next indicates to look for will be deployment milestones, measurable outcomes from simulation-to-robot learning pipelines, and concrete updates on how the DSX-aligned modular and liquid-cooled factory designs roll out across the LG ecosystem.