(LEAD) [APEC 2025] Nvidia to deploy 260,000 GPUs to build AI factories with Samsung, Hyundai in S. Korea
GYEONGJU, South Korea, Oct. 31 (Yonhap) -- Nvidia Corp. plans to deploy up to 260,000 graphics processing units (GPUs) in South Korea in partnership with the government and major companies here to build large-scale artificial intelligence (AI) factories in the country, the U.S. chip giant said Friday.
The initiative will involve the South Korean government and major industry leaders, including Samsung Electronics. Co., SK Group, Hyundai Motor Group and Naver Cloud Corp., that will each establish their own AI computing centers powered by Nvidia's latest Blackwell GPUs.
Under the plan, which Nvidia shared in a preview media briefing on Thursday, the Seoul government will allocate up to 50,000 Nvidia GPUs to support the creation of a national "sovereign AI" platform designed to train large language models (LLMs) optimized for Korean language and industrial applications.
A GPU is a high-performance semiconductor chip designed to handle many calculations simultaneously and is widely used to train and run AI models.
Samsung, SK and Hyundai Motor will each deploy up to 50,000 GPUs, while Naver Cloud will install around 60,000 units, bringing the total to 260,000.
Nvidia said the Blackwell-based infrastructure will boost South Korea's total installed AI GPU capacity from about 65,000 to more than 300,000 units, positioning the country as one of the world's largest hubs for AI computing outside the United States.
Jensen Huang, Nvidia's chief executive officer (CEO), plans to share further details of the initiative on the sidelines of the Asia-Pacific Economic Cooperation (APEC) CEO Summit under way in Gyeongju, the host city of this year's multilateral gathering of regional leaders, later in the day.
Huang arrived in South Korea on Thursday and met with Hyundai Motor Group Executive Chair Euisun Chung and Samsung Electronics Chairman Lee Jae-yong in Seoul.
The Nvidia chief said his visit came at the invitation of President Lee Jae Myung, and said there will be an announcement on details of his company's partnership with South Korean companies.
Financial terms of the arrangements were not disclosed.
Samsung will use the new AI factory to accelerate semiconductor manufacturing through digital twin simulations and smart robotics, employing Nvidia's NeMo, CUDA-X and Omniverse platforms, according to the company.
SK Group said it will design a similar facility to support semiconductor research and cloud-based AI applications, while SK Telecom Co. will provide sovereign industrial cloud services using Nvidia's RTX Pro 6000 Blackwell Server Edition GPUs.
Hyundai Motor will use the 50,000 Blackwell GPUs to train large AI models for autonomous driving, robotics and smart factory operations, with plans to co-invest around US$3 billion with the government to build a "physical AI" ecosystem.
With the Blackwell GPUs, Naver Cloud plans to provide the computing backbone for domestic AI foundation models.
LG Group will also partner with Nvidia in the fields of robotics and medical technology.
In particular, LG and Nvidia plan to advance robotics capabilities and support an AI-driven cancer diagnosis research ecosystem for startups and academia using LG AI Research's large-scale AI model EXAONE.
In addition, Nvidia plans to work with Samsung, SK Telecom, KT Corp., LG Uplus Corp. and Yonsei University to develop AI-native 6G radio access networks (AI RAN), a next-generation wireless system.
odissy@yna.co.kr
(END)
