NHN FactoryX claims top domestic ranking on global supercomputer Top500 list

A GPU cluster operated by NHN FactoryX, the AI infrastructure brand of NHN Cloud, has secured the highest domestic ranking on the global supercomputer Top500 list.

NHN Cloud said Thursday that the core GPU cluster of NHN FactoryX was listed at the top domestic position on the global supercomputer Top500, which was released Monday.

The Top500 is a long-running project in which experts from academic and research institutions in the high-performance computing field have published global supercomputer rankings every June at the International Supercomputing Conference in Germany and every November at the Supercomputing Conference in the United States since 1993. The 67th edition of the rankings was released Monday.

Last year, NHN Cloud was selected as a key operator under a large-scale GPU procurement, construction and operation support project that the Ministry of Science and ICT pursued as part of the government's national mid- to long-term AI infrastructure expansion strategy. Using 7,656 Nvidia B200 GPUs secured through the project, the company built two clusters: NIPA-CL1, comprising 4,080 GPUs across 510 nodes, and NIPA-CL2, comprising 2,040 GPUs across 255 nodes.

Benchmark results showed that NIPA-CL1 recorded a peak measured performance (Rmax) of 137.4 petaflops — a unit representing 1,000 trillion operations per second — placing it first in South Korea and 20th globally. NIPA-CL2 recorded 68.42 petaflops, ranking fourth domestically and 40th globally. NHN Cloud said it has established itself as an AI infrastructure firm operating two of South Korea's top five supercomputers.

NHN Cloud said the results reflect improvements in GPU clustering efficiency built on infrastructure capabilities accumulated since 2023, when it began constructing and operating the Gwangju National AI Data Center. NIPA-CL1 is the largest B200-based GPU cluster in South Korea and features advanced high-density networking technology along with air-conditioning and cooling solutions. The AI-dedicated data center housing the cluster, FactoryX Seoul, employs a 100 percent liquid-cooled GPU cooling system capable of managing a high-density environment of 75 kilowatts per rack.

NHN Cloud described the listing as evidence that South Korea's AI infrastructure capabilities have reached a global level. The company said the government's policy to expand AI computing resources has produced tangible results, laying the groundwork for supplying large-scale AI computing infrastructure domestically in an area that had previously relied heavily on overseas providers.

NHN Cloud also published the NHN FactoryX Technical White Paper, which brings together the GPU cluster optimization expertise and liquid-cooled data center design experience behind the achievement. The company said it aims to share the infrastructure knowledge accumulated by NHN FactoryX with the domestic AI ecosystem and fulfill its role as South Korea's leading AI infrastructure firm.

The white paper covers: a comparison of general-purpose and AI-dedicated data center architectures; GPU cluster configuration and high-speed network interconnect technology; data center power and cooling design and direct-to-chip liquid-cooled infrastructure; the operational structure of a GPUaaS platform using dynamic GPU allocation technology; and case studies on building exascale clusters and enterprise private AI infrastructure.

"This achievement means NHN Cloud has not only played an active role in the national AI infrastructure procurement initiative, but has also firmly established itself as the undisputed No. 1 AI infrastructure operator in South Korea through its GPU cluster construction capabilities," said Kim Dong-hoon, chief executive of NHN Cloud. "We will share these results broadly through initiatives such as the technical white paper and become a leading Korean cloud company that grows together with the domestic AI ecosystem."