NHN Cloud said Wednesday that the core graphics processing unit (GPU) cluster of its artificial intelligence (AI) infrastructure brand "NHN FactoryX" earned the highest domestic ranking on the "Global Supercomputer TOP500" released on June 23.
The "Global Supercomputer TOP500" is a long-running project that has ranked supercomputers worldwide since 1993, with results released each June at the International Supercomputing Conference (ISC) in Germany and each November at the Supercomputing Conference (SC) in the United States.
Last year, NHN Cloud was selected as a major operator in the large-scale "GPU Procurement, Construction, and Operation Support Program" pursued by the Ministry of Science and ICT (MSIT) under the national mid-to-long-term AI infrastructure expansion strategy. Based on 7,656 units of Nvidia's "B200" GPUs secured through the program, the company successfully built two clusters: "NIPA-CL1," with 4,080 units (510 nodes), and "NIPA-CL2," with 2,040 units (255 nodes).
In benchmark results, NIPA-CL1 recorded 137.4 PF (petaflops, a unit measuring 1,000 trillion operations per second) based on maximum achieved performance (Rmax), placing it at the top domestic ranking (20th globally). NIPA-CL2 logged 68.42 PF, ranking fourth domestically (40th globally). With these results, NHN Cloud has established itself as a leading AI infrastructure company operating two of Korea's top five supercomputers.
NHN Cloud maximized GPU clustering efficiency by consolidating the GPU infrastructure capabilities and operational know-how it has accumulated since 2023 through the successful construction and operation of the Gwangju National AI Data Center. NIPA-CL1, which earned the top domestic ranking, is the country's largest B200-based GPU cluster, concentrating all of NHN Cloud's infrastructure technology and capabilities, and is characterized by the application of ultra-high-difficulty networking technology (InfiniBand) and air-conditioning/cooling solutions.
"FactoryX Seoul," the dedicated AI data center where the GPU cluster is built, is recognized for significantly improving infrastructure stability compared with conventional air-cooling systems by applying a 100% liquid-cooling GPU cooling system that controls a high-density environment of 75 kW per rack.
In addition, NHN Cloud recently published the "NHN FactoryX Technical White Paper," which consolidates the GPU cluster optimization know-how and liquid-cooling data center design experience that enabled the 20th and 40th rankings on this Global Supercomputer TOP500.
"This achievement means that NHN Cloud is active at the forefront of the national AI infrastructure procurement project, and that it has firmly established itself as Korea's undisputed No. 1 AI infrastructure operator through its outstanding GPU cluster construction capabilities," NHN Cloud CEO Kim Dong-hoon said. "We will not keep these results merely as an internal achievement, but will widely share NHN FactoryX's technological prowess through measures such as the publication of the technical white paper, becoming Korea's representative cloud company that grows together with the domestic AI ecosystem."
