The joint AI infrastructure project will use digital twins and AI agents to reshape South Korea's semiconductor industry, with completion set for 2027.
NVIDIA and SK Group have announced a partnership to construct one of South Korea's largest AI factories, featuring over 50,000 NVIDIA GPUs. The first phase of the project is scheduled for completion by late 2027, positioning it as a cornerstone of the country's digital transformation efforts.
The AI factory will serve SK subsidiaries including SK hynix and SK Telecom, while also offering GPU-as-a-service to external organizations. This expansion builds on the companies' existing collaboration on high-bandwidth memory and next-generation solutions for NVIDIA's GPU platforms.
NVIDIA CEO Jensen Huang described the AI factory as "a new kind of manufacturing plant" that has emerged in the AI era. He emphasized SK Group's role as a vital memory technology partner in creating advanced computing platforms that drive global AI progress.
SK Group Chairman Chey Tae-Won stated the partnership aims to make AI "the engine of a profound transformation" for Korean industries. The infrastructure is expected to power next-generation developments in memory technology, robotics, digital twins, and intelligent AI agents.
The factory infrastructure will support South Korea's Sovereign AI Foundation Models project, with SK Telecom participating as a model developer. This initiative will provide foundation models for enterprises and researchers working on AI agent development.
SK Telecom plans to build an industrial AI cloud using NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. The initial deployment of over 2,000 GPUs will run NVIDIA Omniverse workloads to support semiconductor manufacturing and digital twin development.
SK hynix is leveraging the infrastructure to accelerate chip design through AI physics and CUDA-X technologies. The company is building autonomous fab digital twins that enable real-time simulation and optimization of manufacturing operations.
SK Telecom is developing a foundation model called A.X. to power AI agents for SK hynix's 40,000 employees. Built with NVIDIA NIM microservices, these agents aim to accelerate productivity across chip development and fabrication processes.