South Korea launches national AI model in tech race with U.S. and China

In a bold move to secure its foothold in the global AI landscape, South Korea is mobilizing its leading corporations and innovative startups to construct a national foundational AI model primarily utilizing domestic technologies. This initiative is aimed at keeping pace with the technological advancements of the U.S. and China. The Ministry of Science and ICT (MSIT) has announced the selection of five consortia tasked with developing the AI models, spearheaded by major players like SK Telecom, which is collaborating with gaming giant Krafton and chip startup Rebellions. Additional teams are being led by other significant firms including LG and Naver. Kim Taeyoon, head of the foundation model office at SK Telecom, emphasized the critical nature of this initiative for South Korea's technological progress. He stated that the nation is committed to laying a robust technical foundation to enhance its competitiveness in the AI sector. "Korea has many entities that would excel at creating a big AI industry. And we could clearly see the possibility that we are very capable of creating a good AI stack," Kim noted, referring to the comprehensive array of technologies that comprise a product. The initiative aims to leverage the strategic advantages of South Korean firms, utilizing their cutting-edge technologies essential for AI development. For instance, SK Hynix produces high-bandwidth memory crucial for AI applications, while Samsung is a key player in the memory market. SK Telecom is also investing in data centers, and Rebellions is focused on designing chips optimized for AI workloads. According to Nick Patience from The Futurum Group, the country's capabilities span the entire AI stack, from hardware to cloud solutions and AI models, bolstered by a thriving community of advanced AI researchers who are actively contributing to academic and patent advancements. However, the initiative acknowledges the complexities of global technology supply chains, as the consortia will still depend on GPUs from NVIDIA, a leading American company, which are pivotal for training AI models. SK Telecom plans to utilize its Titan supercomputer, composed of NVIDIA GPUs, alongside an AI data center in collaboration with Amazon for its model training efforts. SK Telecom is no newcomer to AI, having launched a beta version of its chatbot based on its proprietary large language model, known as 'A.' in 2022. The consortium is targeting a release of its first model by the end of this year, initially focusing on the South Korean market, with potential for global use. The model will be open-source, allowing developers to build upon it with some licensing conditions. Competing against established players like OpenAI and Anthropic, as well as strong offerings from Chinese companies such as Alibaba and DeepSeek, the consortium faces the challenge of creating models that stand out in a crowded marketplace. Kim stated that their aim is to develop a state-of-the-art open-source model that matches the performance of leading tech companies. The implications of this initiative extend beyond national interests, as a robust sovereign AI model could provide significant export opportunities, akin to South Korea's success in the memory chip sector. This would position South Korean AI solutions as viable alternatives to those developed in the U.S. and China. The push for sovereign AI reflects a growing global trend, as nations prioritize the development of AI capabilities within their borders to maintain strategic control. As Patience pointed out, with AI increasingly influencing vital sectors such as healthcare, finance, and defense, countries cannot afford to relinquish control over their digital intelligence to foreign entities.