NHN Cloud Launches Yangpyeong AI Data Center with 7,656 Nvidia B200 GPUs in South Korea
Tags Hardware · Infrastructure · AI
NHN Cloud has begun operating an AI-only data center in Seoul's Yangpyeong district with 7,656 Nvidia B200 GPUs, making it one of South Korea's largest GPU-as-a-service facilities. The facility was built under a Ministry of Science and ICT program with a total budget of 1.46 trillion won. NHN Cloud was selected as the largest operator, responsible for over 1 trillion won of the budget. About 4,000 GPUs are configured into a single cluster for large-scale high-performance computing. A full liquid-cooling system cuts energy use by 15-20% versus air cooling. 6,120 of the 7,656 GPUs are allocated for national-led AI projects; the rest for industry-academia-research users.
Technical significance
South Korea's 7,656 B200 GPU deployment represents a significant national AI infrastructure investment. The 4,000-GPU single cluster configuration enables large-scale model training that was previously only available to hyperscalers. The government-led allocation model (6,120 GPUs for national projects) reflects a strategic approach to AI capacity building that other nations may study.