NVIDIA has announced a landmark collaboration with South Korea’s government and some of the country’s largest industrial players to build out significant AI infrastructure, reinforcing its position as a global leader in accelerated computing. The company will supply hundreds of thousands of its latest-generation GPUs to support “AI factories,” national infrastructure projects and high-performance computing aimed at accelerating manufacturing, mobility, and foundational AI model development.
Under the agreements revealed during a visit to South Korea, Samsung will receive roughly 50,000 NVIDIA GPUs to power a new AI-enabled megafactory focused on semiconductor manufacturing, robotics and device development. SK Group will similarly deploy about 50,000 GPUs to build a manufacturing and cloud infrastructure centred on digital twin technology and agent-based AI tools. Hyundai Motor Group is teaming up with NVIDIA and the government to build an AI-driven mobility and smart-factory ecosystem — again in the range of 50,000 GPUs — aimed at autonomous vehicles, robotics and smart manufacturing. Naver Cloud, meanwhile, will add about 60,000 GPUs as part of its expansion in AI infrastructure and foundation-model development.
The packages combine hardware deployment, software stacks (including NVIDIA’s CUDA-X libraries, Omniverse digital-twin platforms and DRIVE/AGX for mobility), and ecosystem support — signifying a shift from single-vendor sales to deep, strategic co-development. For South Korea, the move is part of a national push to transition from traditional manufacturing to “intelligent production” and position itself among the world’s top AI hubs.
For NVIDIA, the expansion reflects several key strategic trends: moving into growth markets beyond the U.S.-China export conflict, embedding chips in vertical industries rather than just data centres, and locking in long-term compute commitments from industrial and sovereign customers. By aligning with manufacturing, automotive and cloud players, NVIDIA is broadening its role — from AI-training provider to end-to-end ecosystem partner.
There are still questions around delivery timelines, integration complexity and how deeply these AI factories will shift actual operations (versus incremental upgrades). But the scale and ambition are clear: hardware, software and industry are converging under a new paradigm of “physical AI” — real-world systems built on accelerated compute rather than just cloud-based models.
The initiative could accelerate global competition for AI infrastructure as countries and companies rush to stake leadership. For South Korea’s tech giants and the government, this is a bet on turning manufacturing strength into AI capability. For NVIDIA, it’s a bold move to deepen its dominance in the AI compute stack and diversify across sectors.
