Korean AI Services: HyperCLOVA X, Kakao, SKT, Exaone — 2026-09-10

Top developments

Ministry Injects $3.49 Billion for Frontier AI Model

On September 8, the Ministry of Science and ICT announced plans to invest 4.7 trillion won ($3.49 billion) into a project to develop a homegrown "frontier" AI model. This initiative aims to consolidate resources for a single, world-class sovereign model rather than spreading funds across multiple smaller projects. The move is designed to ensure South Korea remains competitive against global AI leaders by focusing on high-performance computing and advanced model architecture.

Naver Cloud Wins Cybersecurity AI Contract

On September 3, Naver Cloud was selected as the final operator for the government’s "Cybersecurity-Specialized AI Foundation Model" development project. The consortium, which includes LG CNS and Korea Electric Power Corporation (KEPCO), beat out a rival bid from SK Telecom. This project is critical for developing sovereign AI capabilities specifically tailored to protect national digital infrastructure and public sector systems. Development is scheduled to begin immediately following the selection.

"AI for All" Consortiums Launch Consumer Agents

SK Telecom, KT, and Kakao have officially begun development of their respective "AI for All" services, targeting 10 million monthly users per platform within one year of launch. The government-backed initiative aims to provide free, unlimited generative AI access to all 52 million citizens via messenger apps and text services. A key requirement for these services is an 80% quota for domestic model usage, ensuring that Korean technology like HyperCLOVA X, Kanana, and A. remains central to the user experience.

National Supercomputer No. 6 Set for December Launch

The National Supercomputer No. 6 is scheduled to begin full operations in December 2026, providing 8,496 GPUs to research institutions. This infrastructure is part of the broader effort to support the training of large language models and scientific AI applications. The deployment addresses growing concerns about compute bottlenecks, as researchers and developers face long wait times for high-performance GPU access.

Local view

Local media outlets are highlighting the tension between rapid policy expansion and practical implementation challenges. Chosun Ilbo notes that while the government has quickly expanded the scope of AI policies, issues regarding market penetration and infrastructure readiness remain significant hurdles for the upcoming National Assembly audits. The Financial News reports on South Korea's ambition to become a global hub for AI safety and security norms, with plans to host international AI organizations in a dedicated campus by the first half of 2027.

Context & numbers

• Total AI Budget: The government's 2026 AI budget has increased to 9.4 trillion won, reflecting a comprehensive strategy covering model development, security, and service deployment.
• Frontier Model Investment: The specific allocation for the new frontier model project is 4.7 trillion won ($3.49 billion).
• User Targets: The "AI for All" program targets 30 million combined users across SKT, KT, and Kakao platforms by 2030, with interim goals of 10 million per platform in the first year.
• GPU Deployment: Over 20,000 high-performance GPUs are planned for deployment across national computing centers by the end of 2027, with Supercomputer No. 6 contributing 8,496 units this year.

On the radar

• National Assembly Audits (October): Lawmakers are preparing to scrutinize the market impact of the "Independent AI Foundation Model" (Dokpamo) project and the utilization rates of the 260,000 GPU deployment plan. Issues regarding child safety in AI chatbots are also expected to be raised.
• International AI Campus: Construction and planning for a campus to house international AI regulatory bodies is ongoing, with a target opening in H1 2027.