Overview
- The Ministry of Science and ICT plans up to 50,000 GPUs for sovereign AI, starting with an initial 13,000-unit deployment led by NAVER Cloud with NHN Cloud and Kakao on the National AI Computing Center.
- Samsung will build a semiconductor AI factory with more than 50,000 GPUs, using Omniverse digital twins and the cuLitho library to accelerate chip manufacturing workflows and logistics.
- Hyundai Motor Group will deploy 50,000 Blackwell GPUs and invest about $3 billion with NVIDIA, establishing an AI Technology Center and a Physical AI Application Center to advance smart factories, robotics and in‑vehicle AI.
- SK Group is developing an AI factory exceeding 50,000 GPUs, with the first phase targeted for completion by late 2027, as SK Telecom launches an industrial cloud using RTX PRO 6000 Blackwell GPUs for digital twins and robotics.
- NAVER Cloud plans to expand to over 60,000 GPUs, while MSIT, NVIDIA and partners including LG AI Research, SK Telecom, NC AI and Upstage will build Korean foundation models using NeMo and open Nemotron datasets.