Naver Cloud Consortium "Successfully Completed AI Semiconductor Farm Construction"

Naver Cloud, acting as the lead organizer, announced on the 6th that the ‘AI Semiconductor Farm Construction and Demonstration’ project, led by the Ministry of Science and ICT and the National IT Industry Promotion Agency, has successfully achieved its first-year goals.

The ‘AI Semiconductor Farm Construction and Demonstration’ project aims to expand the market share of domestically produced AI semiconductors within the Korean data center market to strengthen cloud competitiveness and elevate the technological level of domestic AI semiconductors. It is one of the ‘K-Cloud’ projects overseen by the Ministry of Science and ICT. Spanning three years from May 2023 to November 2025, the project focuses on three core initiatives: building computing infrastructure based on domestic AI semiconductors, establishing and operating a cloud platform, and demonstrating AI application services. Through leading this project, Naver Cloud plans to build a data center capable of high-performance and ultra-low-power computing based on domestic AI semiconductors, thereby generating demand for AI semiconductors and securing global-level competitiveness for its AI-SaaS platform.

Naver Cloud formed a consortium with major domestic CSPs, NHN Cloud and KT Cloud, and collaborates with key domestic AI semiconductor companies such as FuriosaAI, Rebellion, and Sapeon Korea. Additionally, it has established a joint support council with AI service companies including Raonroad, Upstage, Synergy AI, and Nota to demonstrate AI application services in fields such as healthcare and surveillance.

As the project lead, Naver Cloud will build a main NPU computing capacity of 3.5PF (petaflops; 1PF equals one quadrillion operations per second) over three years for AI semiconductor demonstration, oversee detailed execution plans, and manage overall project goals. Participating companies KT Cloud and NHN Cloud will each achieve main NPU capacities of 3.45PF and 10PF respectively, develop cloud software to support the project, and operate cloud services for demonstration support. Furthermore, all three companies plan to build sub NPU computing capacities of 1PF each.

The consortium has set a final goal of achieving a total computing capacity of 19.95PF and demonstrating four application services over three years to build computing infrastructure based on domestic AI semiconductors and to develop AI-SaaS platform solutions. In the first year, they successfully achieved the goal of building 1.1PF capacity, demonstrated one AI application service in the surveillance field, and completed cloud platform analysis and design.

The second-year goals include accumulating 16.95PF capacity to prepare for demonstration through service stability improvements and enhancements, and demonstrating three application services. In the final third year, the plan is to reach a cumulative 19.95PF capacity, demonstrate one additional application service, advance and stabilize the AI Semiconductor Farm and demonstration services, and provide services based on the AI Semiconductor Farm.

Kang Min-seok, Public Business Leader at Naver Cloud, stated, “With the emergence of generative AI based on large-scale language models, the importance of AI semiconductors specialized for AI computing is growing. We have taken a solid first step toward strengthening the competitiveness of domestic AI semiconductors and AI-SaaS.” He added, “We will do our best to create demand for domestic AI semiconductors and AI-SaaS, achieve the final goals, and successfully complete the project.”