Skip to content

South Korea Commits $390M to Boost Sovereign AI Capabilities

South Korea Commits $390M to Boost Sovereign AI Capabilities
Published:

Seoul, South Korea – The South Korean Ministry of Science and ICT launched a ₩530 billion (approximately $390 million) initiative last month to foster the development of sovereign large-scale foundational AI models. This program aims to diminish national reliance on foreign AI technologies and enhance data control within the country, according to ministry statements.

Five local companies were selected to participate in the initial phase: LG AI Research, SK Telecom, Naver Cloud, NC AI, and the startup Upstage. The government plans to conduct biannual progress reviews, discontinuing funding for underperforming entities and consolidating support for frontrunners, ultimately aiming to retain two primary leaders for the national AI drive. NC AI declined to comment on its participation.

LG AI Research, an R&D unit of LG Group, is developing Exaone 4.0, a hybrid reasoning AI model. This iteration integrates broad language processing with advanced reasoning features, building on its Exaone Deep predecessor. Co-head Honglak Lee stated that LG focuses on refining data from diverse industries, including biotech and manufacturing, to deliver practical value rather than solely pursuing model scale. The company aims to improve models via API-generated real-world user data, emphasizing efficiency over massive GPU cluster investments.

SK Telecom introduced its A.X large language model in July, following its A. personal AI agent service launched in late 2023. SK Telecom reports that A.X 4.0 processes Korean inputs approximately 33% more efficiently than GPT-4o. The company has also open-sourced its A.X 3.1 models. Taeyoon Kim, head of the foundation model office at SK Telecom, highlighted the company's access to telecom network data, including navigation and taxi-hailing information, as a key advantage. SK Telecom is also investing in AI infrastructure, including South Korea’s largest GPU-based service, GPUaaS, and a new hyperscale AI data center with AWS.

Naver Cloud, the cloud services arm of Naver, unveiled its HyperCLOVA X model in 2023, along with an AI chatbot, CLOVA X, and a generative AI search engine, Cue. A Naver spokesperson stated that the company operates an "AI full stack," encompassing model development, data centers, cloud services, and applications. Naver embeds its AI into core services such as search, shopping, and maps, leveraging its real-world data for recommendations. The company prioritizes model sophistication and capital for scaling, rather than simply chasing size.

Upstage, the sole startup in the cohort, launched its Solar Pro 2 model in July. Soon-il Kwon, executive vice president at Upstage, noted that Solar Pro 2, with 31 billion parameters, has been recognized as a frontier model by Artificial Analysis. Kwon reported that Solar Pro 2 has outperformed global models on major Korean benchmarks, with Upstage aiming to achieve 105% of the global standard for Korean language performance through this project. The company intends to focus on industry-specific models for sectors like finance, law, and medicine.

More in Live

See all

More from Industrial Intelligence Daily

See all

From our partners