Fuelled by 50,000 NVIDIA GPUs, SK Group’s new AI factory promises to reshape Korea’s tech landscape, pushing breakthroughs in memory, robotics and next-generation AI innovation.
SK Group and NVIDIA have partnered to drive breakthroughs in high-bandwidth memory (HBM), semiconductor manufacturing and artificial intelligence infrastructure in South Korea. The partnership will result in the creation of one of the country’s largest AI factories, powered by over 50,000 NVIDIA GPUs.
The project, expected to complete its first phase by late 2027, will serve SK Group subsidiaries, such as SK hynix and SK Telecom, as well as external clients through a GPU-as-a-service model. The facility aims to accelerate research, digital transformation and AI innovation across industries in South Korea.
Furthermore, under the collaboration, SK hynix will co-develop next-generation memory technologies for NVIDIA GPUs, enhancing performance for AI and data-intensive applications. The company is using NVIDIA’s CUDA-X technologies, including the PhysicsNeMo framework, to speed up semiconductor design through AI physics and simulation.
SK hynix is also building autonomous fab digital twins using NVIDIA Omniverse libraries and RTX PRO Servers to optimise manufacturing efficiency and progress toward self-optimising production.
Meanwhile, SK Telecom will build an industrial AI cloud in Asia, using more than 2000 NVIDIA RTX PRO 6000 Blackwell GPUs. This platform will support digital twin and robotics innovation for manufacturing firms, startups and government agencies.
NVIDIA CEO Jensen Huang described the venture as “a new kind of manufacturing plant — the AI factory.” At the same time, SK Group Chairman Chey Tae-won said the collaboration will help industries “transcend traditional limits of scale, speed and precision.”
