Nvidia VP for global AI cloud and infrastructure ecosystem speaks at online media briefing
HyperCLOVA X to be upgraded with Nemotron
Plans extend beyond Sejong data center to global expansion
'South Korea's manufacturing strength makes it ideal for physical AI growth'
![Nvidia CEO Jensen Huang (right), SK Group Chairman Chey Tae-won (first from right), LG Group Chairman Koo Kwang-mo (second from left) and Naver Chairman Lee Hae-jin (first from left) meet at a samgyeopsal restaurant called "Hyeongnim Jeoyo" near Hongdae in Mapo-gu, Seoul, on June 5 for what has been dubbed a "samso" — short for samgyeopsal and soju — dinner. [Yonhap]](https://wimg.heraldcorp.com/news/cms/2026/06/08/news-p.v1.20260605.9b9a07cfc0f34498903b879e5c817b97_P1.jpg)
The results of the much-discussed dinner between Naver Chairman Lee Hae-jin and Nvidia CEO Jensen Huang are coming into focus. Naver has agreed to deepen its AI infrastructure partnership with Nvidia.
The collaboration will use Nvidia's DSX platform — an AI factory design architecture — to address domestic and international AI cloud customer demand while advancing Naver's sovereign AI ambitions. Naver is looking beyond sovereign AI development to broader expansion of its AI data center business.
Raj Mirpuri, Nvidia's vice president for global AI cloud and infrastructure ecosystem, said Sunday at an online media briefing that "Naver is building a gigawatt-scale AI factory based on Nvidia DSX to strengthen its agentic AI and physical AI capabilities."
Nvidia unveiled the DSX platform at GTC Taipei 2026 earlier this month. The full-stack AI factory infrastructure solution covers everything from data center construction and power supply to platform deployment and token production.
"Nvidia has a very deep interest in the Korean ecosystem," Mirpuri said. "Naver has very strong cloud capabilities, so this will be an excellent combination of two partners."
Naver will also strengthen its sovereign AI capabilities on the Nvidia AI platform. The company already operates HyperCLOVA X, its large-scale AI language model, and is upgrading it using Nvidia's open large language model Nemotron 3 Ultra.
"Naver has AI cloud infrastructure, Korean-language AI expertise and strong local customer access," Mirpuri said. "Naver is the first Korean company to join the Nvidia Nemotron Alliance and is contributing to open model development across pre-training, post-training and reinforcement learning."
Naver is also using the deepened partnership as a springboard into physical AI. The company is building a "Seoul World Model" based on Cosmos 3, a foundation model for physical AI.
Nvidia pursued the collaboration on the strength of Naver's dominant position in the domestic market.
"Naver and we have worked together in the AI computing platform space for years, and even before that," Mirpuri said. "Naver has outstanding technical talent and remarkable influence locally."
Naver also plans to expand its data center footprint as part of the alliance. The company currently operates a data center in Sejong.
"Naver has plans to expand its data center capacity," Mirpuri said. "It has been designed with not only domestic capacity in mind but also global expansion."
Nvidia also highlighted South Korea's growing AI market, singling it out as a high-potential market for physical AI. "We are living in the era of generative AI and agentic AI," Mirpuri said. "South Korea is a manufacturing powerhouse, so every factory and every sensor on the ground will become a driver of physical AI growth."
jeongwan@heraldcorp.com
