NHN Cloud Selected as Largest Provider in Government GPU Procurement Project

Utilizing 1 Trillion KRW... Securing 7,656 GPUs
Introducing Water-Cooling System for Effective GPU Cooling

NHN Cloud announced on July 29 that it has been selected as the largest provider in the 'AI Computing Resource Utilization Foundation Enhancement (GPU Procurement, Construction, and Operation Support) Project' promoted by the Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA).

This project is designed to select providers capable of procuring, constructing, and operating large-scale advanced Graphics Processing Units (GPUs). It is being carried out through public-private cooperation to expand the foundation for global AI competitiveness at the national level. The Ministry of Science and ICT announced the final providers the previous day, selecting three companies: Naver Cloud, NHN Cloud, and Kakao.

Among the three companies selected as project participants, NHN Cloud was chosen as the provider responsible for deploying the largest number of GPUs. NHN Cloud will use more than 1 trillion KRW out of the total 1.46 trillion KRW budget to supply GPU services. Out of the total 13,000 GPUs secured for this project, NHN Cloud will deploy the largest share with 7,656 units of NVIDIA's B200, accounting for approximately 75% of the total latest B200 models.

In addition, as the lead provider, NHN Cloud will establish and provide a 'GPU Infrastructure Integrated Portal' that can be jointly used by all three companies.

NHN Cloud plans to implement a water-cooling system, which is effective for cooling high-performance GPUs that generate significant heat. NHN Cloud explained that it is particularly significant that, for the first time in Korea, more than 4,000 GPUs will be configured in a single cluster with a water-cooling system, thereby demonstrating the technical capability required for building large-scale GPU infrastructure.

The colocation data center will be built to global standards, establishing a high-density power base of 75kW per rack. The focus is on providing a stable and efficient GPU infrastructure.

The data center is scheduled to complete system construction by the end of this year and launch beta services on January 1 next year. NHN Cloud plans to begin pilot operations in January next year and, after meeting project requirements, officially launch the service in mid-March next year.

Kim Donghoon, CEO of NHN Cloud, stated, "This selection is a recognition of NHN Cloud's experience as the first in Asia to introduce the latest GPUs, build the Gwangju National AI Data Center, and operate GPU as a Service (GPUaaS) businesses. It will serve as a stepping stone for NHN Cloud to secure a leading position in large-scale GPU infrastructure projects. As a leading domestic cloud service provider (CSP), NHN Cloud will fulfill its role in rapidly expanding the AI computing infrastructure promoted by the Ministry of Science and ICT and NIPA for the nation's future."