NVIDIA, NAVER and SK Telecom Announce Plans for Gigawatt‑Scale Sovereign AI Factories

The partnerships aim to move AI from research to production using NVIDIA’s DSX full‑stack blueprint, with rollout dependent on power procurement and undisclosed investment and capacity plans.

Overview

On June 7, 2026, NVIDIA issued coordinated announcements with NAVER and SK Telecom that commit both Korean firms to build DSX‑based sovereign AI clouds starting with NAVER’s 55 MW expansion at GAK Sejong and SK Telecom’s first AI factory planned for 2027.
Both companies said they will use NVIDIA’s DSX platform—a codesigned stack of chips, systems and software intended to lower token production costs and speed time to production for large AI workloads.
NAVER will advance its HyperCLOVA X and develop a Seoul World Model by fine‑tuning Nemotron models and joining NVIDIA’s Nemotron Coalition to contribute datasets and training work.
SK Telecom plans a gigawatt‑scale AI Cloud to power training, inference and agentic physical AI for industry and enterprise and will pursue joint research with SK Group and NVIDIA on next‑generation AI factory architecture.
Key implementation details such as total investment, GPU counts and precise schedules remain undisclosed and both NVIDIA executives and partners say progress is contingent on securing sufficient power and other operational approvals.