Naver and Nvidia Forge AI Infrastructure Alliance... Building a Gigawatt-Scale AI Factory (Comprehensive)

Building a 1GW-Scale AI Factory... Four Times the Capacity of 'GAK Sejong' Data Center
Leveraging the Nvidia DSX Platform... Enhancing LLMs with Nemotron
First Korean Company to Join the 'Nemotron Alliance'
Collaboration in Physical AI... Constructing the 'Seoul World Model'

Naver is joining forces with Nvidia to build a gigawatt (GW)-scale artificial intelligence (AI) factory. The two companies plan to collaborate in a wide range of areas, including AI data centers, sovereign AI models, large language models (LLMs), and physical AI.

Raj Mirpuri, Vice President of Nvidia's Global AI Cloud and Infrastructure Ecosystem division, announced at a media briefing on June 8, 2026, "Nvidia and Naver have agreed to expand sovereign infrastructure to GW scale through the Nvidia DSX platform." On the same day, Naver also disclosed, "In response to the rapidly increasing global demand for AI infrastructure, we intend to jointly build and operate a large-scale global AI factory in strategic partnership with Nvidia."

Jensen Huang, CEO of NVIDIA, is taking a commemorative photo after a samgyeopsal meeting at a barbecue restaurant in Seogyo-dong, Mapo-gu, Seoul, on June 5, together with Chey Tae-won, Chairman of SK, Koo Kwang-mo, Chairman of LG, and Lee Hae-jin, Chairman of Naver. 2026.06.05 Photo by Dongju Yoon

An AI factory is a cutting-edge data center that serves as core infrastructure for model training, post-training, and inference operations, and is operated much like a 'factory' for mass-training AI models.

Initially, the two companies will begin by establishing a 55-megawatt (MW) AI factory in 2027, with plans to build up to GW-scale AI factories. Naver will use its 'GAK Sejong' data center as a forward base. Starting with the 55MW launch in the first half of 2027, Naver aims to expand its overseas infrastructure to 100MW within the same year and to 200MW by 2028.

Through these efforts, the two companies plan to establish a GW-scale AI factory infrastructure. 1GW is approximately four times the maximum capacity of Naver's largest domestic data center, 'GAK Sejong,' and can accommodate several hundred thousand of Nvidia's latest graphics processing units (GPUs) simultaneously.

They also intend to strengthen their technological collaboration. Naver's capabilities in building and operating its own GPU clusters will be combined with Nvidia's DSX platform. DSX is Nvidia's integrated platform that supports the design, deployment, and operation of AI factories. It covers all aspects of AI infrastructure, including data center construction and power supply, accelerated computing systems, software, facility layout, and token production efficiency. The aim is to improve both the operational efficiency and business potential of data centers.

The companies will also collaborate in the physical AI sector. Naver plans to build a 'Seoul World Model' on Nvidia's world foundation model 'Cosmos,' utilizing its proprietary spatial modeling and Street View data. Through this, they will pursue next-generation technological cooperation in the field of spatial intelligence.

On the 1st, at NVIDIA GTC Taipei 2026 held in Taipei, Taiwan, Jensen Huang, CEO of NVIDIA, introduced Naver Cloud as a key artificial intelligence (AI) native cloud partner. Naver Cloud

Recently, Naver became the first Korean company to join the 'Nvidia Nemotron Alliance.' The Nemotron Alliance, led by Nvidia, brings together 12 global AI companies, including Cursor, Mistral AI, and Perplexity. Naver contributes to the development of open models through pre-training, post-training, and reinforcement learning.

Naver is also working to enhance the performance of its own LLM through Nvidia's open foundation model, Nemotron. Naver is upgrading the performance of 'HyperCLOVA X' by adding its proprietary data to Nvidia's 'Nemotron 3 Ultra' open model, which has 550 billion parameters.

Meanwhile, Lee Hae-jin, Chairman of the Naver Board, and Jensen Huang, CEO of Nvidia, will hold a meeting at Naver's 1784 headquarters in Seongnam, Gyeonggi Province, later this afternoon to discuss the concrete roadmap for the projects they are pursuing and detailed cooperation plans for joint entry into the global market. The two companies have already agreed on the direction of expanding their AI infrastructure ecosystem business beyond the Asia-Pacific region into Europe and the Middle East.

Jensen Huang, CEO of Nvidia, said, "Through Nvidia DSX, we will support Korea in expanding its sovereign intelligence infrastructure for the agentic era, from AI agents to AI factories and physical AI."

Lee Hae-jin, Chairman of the Naver Board, commented, "This alliance is highly encouraging as it enables each region and country around the world to establish their own sovereign AI capabilities through concrete alternatives. This partnership is significant in that it provides Naver with an opportunity to take its technological infrastructure competitiveness to the next level in the global market."

This content was produced with the assistance of AI translation services.