SK Telecom (NYSE: SKM) today announced that the SKT Consortium has been selected as a core team for the ‘Proprietary AI Foundation Model’ project led by the Ministry of Science and ICT (MSIT).

As part of this initiative, the SKT Consortium aims to develop a full-stack artificial intelligence (AI) platform based on proprietary technologies spanning semiconductors, models, data and services. A significant aspect of this effort will include open-sourcing these capabilities to empower domestic companies and accelerate Korea’s AI innovation ecosystem.

The SKT consortium brings together industry leaders and research institutions across Korea to establish a solid foundation for proprietary AI advancement. Kim Taeyoon, Head of the Foundation Model Office at SK Telecom, will conduct core research on AI technologies. The main participating companies and research institutions include:

•  Krafton: Korea’s largest listed game company, known for iconic game franchises.

•  42dot: A leader in mobility AI solutions aimed at advancing autonomous technologies.

•  Rebellions: A top-tier manufacturer of AI inference chips (neural processing units, or NPUs).

•  Liner: A specialist in AI-driven knowledge and search agent technologies.

•  Selectstar: A provider of data reliability technologies critical to AI model development.

•  Seoul National University (SNU): Professor Kim Gunhee, Professor Yoon Sungroh, Professor Hwang Seungwon, Professor Do Jaeyoung

•  Korea Advanced Institute of Science and Technology (KAIST): Professor Lee Kimin

In addition, Professor Lee Kangwook and Professor Dimitris Papailiopoulos from the University of Wisconsin-Madison have joined as key researchers in this project, further strengthening the consortium’s AI research capabilities.

The SKT Consortium has demonstrated its expertise in the AI field, with over 800 research papers, 736 patents, and 270 open-source projects.

Image: Full-stack AI concept map of SKT Consortium

SK Telecom is embarking on the development of next-generation large-scale AI models through the Proprietary AI Foundation Model project. The company plans to incorporate ‘omni-modal’ technology, enabling the integrated processing of not only text, but also images, speech and video data.

Since 2018, SK Telecom has been developing its own large language model (LLM), A.X. This year, the company has sequentially released two versions (standard and light) of the A.X 4.0 based on large-scale continual pre-training (CPT), as well as two versions (standard and light) of the A.X 3.1 built entirely from scratch. These models deliver performance on par with GPT-4o and demonstrate world-class ability in understanding Korean context.

To date, SK Telecom has trained A.X models on its own TITAN supercomputer. Moving forward, the SKT Consortium will provide large-scale computing resources independently to conduct advanced research and development. In addition, consortium member Rebellions will leverage its domestically developed NPUs to optimize high-performance, energy-efficient AI services.

The consortium’s ultimate objective is to make AI agents accessible and convenient for all Koreans, while leading AI innovation across key industries such as office work, manufacturing, automotive, gaming and robotics.

“With our proven technological capabilities and operational expertise, we will deliver the highest-quality Korean-style proprietary AI foundation model to empower AI for daily life in Korea,” said Kim Jiwon, Head of AI Model Lab at SK Telecom.