NVIDIA is cementing its position as the backbone of sovereign AI infrastructure by building dedicated 'AI factories' with major industrial conglomerates across geographies, a strategy that transforms the chipmaker from pure-play supplier into infrastructure architect. This week, NVIDIA announced parallel collaborations with LG Group and South Korea's Doosan Group to construct purpose-built compute facilities focused on physical AI applications—robotics, autonomous mobility systems, and cloud GPU services. Simultaneously, the company is operationalizing the UK's 2024 sovereign AI pledge from Prime Minister Keir Starmer and NVIDIA CEO Jensen Huang by showcasing real deployments at London Tech Week. The move signals NVIDIA's recognition that governments and enterprises increasingly demand localized, vertically integrated compute ecosystems rather than simply purchasing chips off-the-shelf.
The LG and Doosan factories represent concrete manifestations of 'physical AI' infrastructure—not abstract cloud compute, but facilities explicitly engineered to train and validate robotics, autonomous vehicle systems, and industrial automation. LG's factory will support its robotics divisions and data center GPU services; Doosan's will anchor work across Doosan Robotics, Doosan Bobcat autonomous equipment, and Doosan Enerbility's energy systems. Both partnerships leverage NVIDIA's full stack—GPUs, CUDA software, Isaac robotics simulation platform, and Omniverse digital twin tools—creating ecosystem lock-in. While NVIDIA has not disclosed specific investment figures, facility timelines, or revenue-sharing terms, the industrial scale of these partnerships with Fortune 500-level conglomerates suggests multi-year, potentially multi-billion-dollar commitments. These arrangements essentially position NVIDIA as both technology enabler and infrastructure co-developer, distinguishing the company from competitors like AMD, which remain primarily chip vendors without comparable end-to-end factory partnerships.
The sovereign AI angle carries geopolitical weight: the UK, South Korea, and implicitly other allied nations are betting that NVIDIA-powered regional factories prevent dependency on U.S. cloud monopolies while strengthening domestic hardware and robotics champions. South Korea's choice is particularly significant, given its existing strengths in semiconductor manufacturing, gaming (NVIDIA's RTX Spark targeting Korean PC bangs), and robotics innovation. However, the strategy also raises competitive questions. AMD and emerging chipmakers have no announced comparable factory programs, potentially ceding the 'sovereign infrastructure' narrative to NVIDIA. Industry analysts caution that over-reliance on a single vendor—however dominant—risks recreating different forms of dependency. The real test will come when these factories begin shipping products: whether LG robots and Doosan autonomous systems prove commercially viable, and whether NVIDIA's contractual arrangements with partners include sufficient profit-sharing to avoid resentment or hedging toward alternative architectures.