Naver Cloud Consortium Launches Development of AI Model Specialized for Domestic Security Environment
The Naver Cloud Consortium announced that it has been selected as the final business operator for the government's 'Specialized Artificial Intelligence Foundation Model Development (Cybersecurity Sector)' project and has begun full-scale development.
Through this project, the consortium, consisting of a total of 33 companies, academic institutions, research organizations, and public agencies, plans to build an AI model optimized for Korea's security environment, and advance the competitiveness of domestic security AI by conducting verification and enhancement at national critical facilities and major industrial sites.
Notably, the consortium has been proactively deploying 4,000 GPUs even before the official project launch to conduct pre-training of security-specialized AI models. In addition to government-provided capacity (256 units), the structure involves deploying Naver Cloud's 4,000 B200 units and LG's 256 H200 units. Based on this comprehensive preemptive investment, the consortium plans to significantly shorten development time, while concentrating the secured time on enhancing model performance and conducting on-site demonstrations to maximize completion quality.
Drawing on this large-scale infrastructure, the consortium will build a massive 700B-class MoE (Mixture of Experts) security-specialized model based on HyperCLOVA X and Exaone. Two proprietary models—one reinforcing defense capabilities (HyperCLOVA X) and another reinforcing attack capabilities (Exaone)—will be developed in parallel to secure comprehensive capabilities for the entire attack-defense cycle. The consortium plans to rigorously verify objective performance using two internationally recognized benchmarks.
Additionally, the high-quality real data used for model training will be approximately 830TB in scale, which is difficult for a single institution to secure. This encompasses operational data from Naver Cloud and LG CNS, as well as core data from national infrastructure and major industries including Korea Electric Power Corporation KDN, Korea Hydro & Nuclear Power, Financial Security Institute, KISTI, and LG U+. Data refining technology spanning from threat intelligence collection through labeling, standardization, and validation will be applied to precisely train the model on actual threat environments.
The model will be continuously enhanced through demonstrations at national critical facilities and major industrial sites. On-site demonstrations will be conducted across seven major sectors including power, finance, science and technology, telecommunications, semiconductors, defense industry, and aerospace. The consortium aims to complete operational expertise specialized for 'closed networks' to enable stable operation even in secure environments with restricted external network access.
Kim Yu-won, CEO of Naver Cloud, stated, "The project award reflects comprehensive recognition of the large-scale AI infrastructure, data, and security expertise possessed by the Naver Cloud Consortium. We will undertake this project with a sense of responsibility to create AI that protects national critical infrastructure through our technology, and will build a national security AI model and ecosystem that represents the Republic of Korea."