Only 14% of Domestic AI Data Centers Achieve Top Grade Under Korean Standards

Korea Artificial Intelligence Industry Association Publishes Research Report on "Establishment of Domestic AI Data Center Evaluation Criteria"

Mega Data and LG Uplus Excel Across All Criteria

Urgent Need for Policy Support and Regulatory Relaxation

As a result of applying the "Korean evaluation criteria," which reflect policy factors such as power grid capacity and site suitability, only 6 out of 42 domestic AI data centers subject to the evaluation exceeded the highest standards. With explosive demand for AI data centers expected, there is an urgent need for policy support and regulatory relaxation moving forward.

An aerial view of the Megadata Data Center in Wonju, Gangwon Province, which received the highest grade (Green) in the "Korean AI Data Center Evaluation" in recognition of its dual power supply route securing and natural disaster risk minimization capabilities.

According to the industry on April 8, the Korea Artificial Intelligence Industry Association recently published a research report titled "Establishment of Domestic AI Data Center Evaluation Criteria." This evaluation was an attempt to address the limitations of widely used international standards such as "TIA-942," which, while valid for assessing technical completeness, have been criticized for failing to reflect Korea’s unique policy environment—such as the structure of the power grid and the permitting process.

The research team developed a new "Korean Multi-Criteria Evaluation Framework," which supplements technology-focused assessments by quantitatively scoring power infrastructure capacity, site and risk factors, permitting environment, and project feasibility, with a maximum score of 205 points.

Based on a strict application of these criteria to 42 business plans, only 6 (14.3%) met the most rigorous "Strict" standard and received the highest "Green" grade. Even when the criteria were relaxed to "Moderate" (12 sites) or "Flexible" (21 sites), a majority or substantial portion of the projects still urgently needed improvements, particularly in areas such as power infrastructure and site suitability.

The only data centers to earn a "Green" rating under all three criteria were Mega Data in Wonju, Gangwon-do—which secured dual power supply paths, minimized natural disaster risks, and ensured a stable construction schedule—and LG Uplus in Paju, Gyeonggi-do, recognized for its corporate reliability and technology certifications.

These findings highlight that, despite the explosive growth in demand for AI data centers, structural issues remain. According to consulting firm McKinsey, the global demand for AI data centers is expected to reach 156 GW by 2030 due to the spread of large language models (LLM), physical AI, and autonomous driving, with AI workloads accounting for 70% of total demand. However, at the current pace of construction, it will be difficult to keep up with this demand.

The domestic market is also projected to grow at a steep annual rate of 26.6%, with 290 applications for data center power usage submitted in just 11 months. Nevertheless, supply of AI-dedicated data centers—which require massive power and cooling facilities for large-scale GPU operations—remains severely inadequate.

Professor Yoo Byungjun of the Seoul National University AI Research Institute, who led the study, emphasized, "AI data centers are not just IT facilities, but strategic assets that determine national digital sovereignty and industrial competitiveness. To meet future demand, the government must quickly implement bold policy support and regulatory relaxation, including preemptive expansion of power infrastructure."

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