SK Telecom announced on April 10 that it will collaborate on the AI Data Center (AIDC) business together with the British semiconductor design company Arm and the domestic AI semiconductor startup Rebellion.


SK Telecom announced on the 10th that it will collaborate on the AI data center (AIDC) business together with the UK semiconductor design company Arm and the domestic artificial intelligence (AI) semiconductor startup Rebellion. Photo by SK Telecom

SK Telecom announced on the 10th that it will collaborate on the AI data center (AIDC) business together with the UK semiconductor design company Arm and the domestic artificial intelligence (AI) semiconductor startup Rebellion. Photo by SK Telecom

View original image

SK Telecom signed a strategic partnership (MOU) with Arm and Rebellion the previous day to drive innovation in next-generation AI infrastructure. Under this agreement, Arm's newly launched "Arm AGI Central Processing Unit (CPU)" and Rebellion's "RebelCard," which is scheduled for release in the third quarter of this year, will be integrated into servers to jointly develop an AI inference optimization solution and test it at SK Telecom's AIDC. The Arm AGI CPU is the first data center processor that Arm has directly produced and is optimized for AI inference services. The RebelCard is a neural processing unit (NPU) specialized for large-scale AI inference.


According to SK Telecom, "By installing both chips in a single server, the CPU can handle general-purpose computing such as data processing and system operations, while the NPU is dedicated to AI inference operations. This can improve power efficiency and reduce operating costs," adding, "This architecture is an efficient server structure for data centers running large-scale AI services."


The background of this collaboration lies in the paradigm shift within the AI industry. The paradigm is shifting from training AI models to inference, which provides actual services. As a result, the core challenge of AI infrastructure is evolving from providing massive computing power for training to delivering AI services quickly and affordably with less power consumption. The industry is paying attention to NPUs—dedicated chips specialized for inference—as a new alternative.


SK Telecom is considering operating its proprietary AI foundation model, A.Dot X K1 (A.X K1), on servers equipped with this solution. Through this collaboration, SK Telecom aims to secure a low-power, high-efficiency AI inference infrastructure and further enhance the competitiveness of its AIDC business.



Lee Jaesin, Head of AI Business Development at SK Telecom, said, "By providing a full package that combines the inference-optimized A.X K1, we will strengthen the competitiveness of our AIDC."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing