South Korea Launches Sovereign AI Drive with Five Homegrown LLM Leaders

Seoul has unveiled a sovereign AI program that funds five domestic firms—LG AI Research, SK Telecom, Naver Cloud, NC AI and Upstage—to build large language models optimized for Korean language and culture. The initiative aims to reduce reliance on foreign AI services, improve data security, and position South Korea to compete with global players such as OpenAI and Google. Each company brings a distinct strategy, from hybrid reasoning models and telecom‑driven data pipelines to full‑stack cloud ecosystems and cost‑effective startup solutions.

Government Backing for a National AI Ecosystem

South Korea’s Ministry of Science and ICT announced a multi‑year sovereign AI initiative that allocates substantial public funding to five selected companies. The program is designed to foster homegrown large language models (LLMs) that understand Korean language nuances and cultural context, while also strengthening national data security and reducing dependence on foreign AI platforms.

LG AI Research: Hybrid Reasoning and Efficiency

LG AI Research, the research arm of the LG Group, is developing the Exaone 4.0 model, a hybrid reasoning system that blends broad language processing with advanced reasoning capabilities. Rather than pursuing raw scale, LG focuses on efficiency, extracting maximum performance from each chip and refining data before model training. The company plans to deliver its models through APIs, using real‑world user interactions to iteratively improve performance and generate economic value.

SK Telecom: Telecom‑Powered AI Services

SK Telecom leverages its extensive telecom infrastructure to power its AI initiatives. The firm recently released the A.X 4.0 large language model, built on an open‑source foundation from Alibaba Cloud. SK claims its model processes Korean inputs more efficiently than leading foreign alternatives. The company’s consumer‑facing AI agent, known as A., provides features such as call summarization and auto‑generated notes, and has attracted millions of users. SK Telecom also invests in GPU‑as‑a‑service and is constructing a hyperscale AI data center in partnership with AWS.

Naver Cloud: Full‑Stack AI Integration

Naver Cloud, the cloud arm of South Korea’s leading internet company, offers the HyperCLOVA X model and a suite of AI‑driven products, including a chatbot and a generative‑AI search engine. Naver emphasizes its “AI full stack,” covering model development, data center operations, cloud services, and consumer applications. By embedding AI across search, shopping, maps and finance, Naver aims to act as a connector that integrates legacy systems and siloed services.

NC AI: Silent Participant

NC AI was selected for the sovereign AI cohort but declined to comment on its specific plans.

Upstage: Startup Innovation with Cost‑Effective Models

Upstage, the sole startup in the cohort, introduced the Solar Pro 2 model, a 31‑billion‑parameter LLM recognized as a frontier model. Despite its relatively modest size, Solar Pro 2 outperforms larger global models on Korean benchmarks and offers a cost‑effective solution for businesses. The company focuses on industry‑specific applications in finance, law and medicine, and seeks to build an AI‑native startup ecosystem.

Strategic Goals and Review Process

The government will assess progress every six months, trimming underperforming participants and continuing support for the most promising projects until only two remain as leaders of the national AI effort. By combining diverse strengths—from data‑rich telecom networks to cloud‑scale infrastructure and agile startup innovation—South Korea aims to create competitive, locally tuned AI capabilities that can stand alongside global heavyweights.