NVIDIA and LG Group Sign AI Factory Partnership Covering Robotics, ADAS, and Sovereign AI
NVIDIA and LG Group announced a sweeping AI factory collaboration on June 7, spanning six LG subsidiaries and four technical work streams from physical AI to autonomous driving. The deal ties Korea's largest conglomerate to NVIDIA's Blackwell infrastructure at a moment when sovereign AI buildout is accelerating across Asia.
NVIDIA and LG Group signed a broad AI factory partnership on June 7, binding together six LG subsidiaries under a shared infrastructure roadmap that runs from home robots to sovereign AI models. Jensen Huang announced the deal during his Seoul visit, part of a string of Korean partnerships that also included SK Hynix, SK Telecom, Naver, Hyundai, and Doosan.
The collaboration has four distinct work streams. Physical AI and robotics is the most visible: LG’s CLoiD home robot, unveiled at CES 2026, will train inside NVIDIA Isaac Sim digital twins and run on the GR00T foundation model. LG Electronics intends to put NVIDIA hardware inside a wider lineup of appliances and home automation products, using Isaac for simulation and validation before physical manufacturing.
The second stream is ADAS. LG Innotek — which supplies camera modules and sensor arrays to global automakers — will integrate NVIDIA DRIVE Hyperion into its autonomous driving component stack. Third, LG CNS and LG Uplus are building out liquid-cooled AI data centers aligned with NVIDIA’s DSX architecture, giving the group a proprietary compute backbone to train and serve large models without depending on third-party cloud providers.
The fourth stream is the most strategically significant: LG AI Research and NVIDIA will jointly develop EXAONE, South Korea’s leading open sovereign AI model, on Blackwell GPUs using the NeMo training framework. Sovereign AI — the push by governments and national conglomerates to build frontier models on domestic infrastructure — has become a defining theme across the Asia-Pacific in 2026, and NVIDIA is positioning itself as the default platform partner.
No financial terms were disclosed. But the breadth of the deal — six subsidiaries, four work streams, alignment with government-backed sovereign AI goals — makes it one of the more substantive corporate AI partnerships announced this year. LG is not buying GPUs for one use case; it is restructuring its AI compute strategy around NVIDIA’s full stack from edge inference to datacenter training.
For developers, the EXAONE angle is worth watching. An open sovereign model trained on Blackwell hardware with a major conglomerate behind it creates a credible Korean-language and multimodal alternative to the US frontier labs. NVIDIA’s willingness to contribute NeMo and simulation infrastructure suggests it sees Korea’s sovereign AI market as a long-term beachhead worth supporting at cost.