0
SK Group Chairman Chey Tae-won (third from left), Nvidia CEO Jensen Huang (second from left), SK Hynix CEO Kwak Noh-jung (left), and SKT CEO Jung Jae-heon take a photo together during a 'Kkanbu meeting' held in Samseong-dong, Seoul, on the 8th. / Photo by Reporter Song Ui-ju

SK Telecom is set to embark on building AI infrastructure targeting the global market in partnership with Nvidia. To this end, SKT will join the 'Nvidia Cloud Partner' program, a global partner ecosystem program that provides high-performance cloud AI services utilizing Nvidia's AI infrastructure and software.

According to SKT on the 8th, the company has agreed to pursue a 'full-stack AI cloud' cooperation with Nvidia encompassing everything from chips to data center operations based on the Nvidia DSX platform. For this purpose, the company plans to scale up 'AI factories'—data centers specialized for AI workloads—targeting a gigawatt-scale (GW) capacity.

The Nvidia DSX platform is a platform that defines how AI factories are designed, built, and optimized across the entire full-stack, ranging from chips and systems to infrastructure software, facilities, and partner technologies.

During a meeting held in Taiwan on the 1st, SK Group Chairman Chey Tae-won and Nvidia CEO Jensen Huang reviewed the AI infrastructure roadmap envisioned by both companies and agreed to push forward with group-wide cooperation. SKT, the primary executing entity of this cooperation, shared that it plans to pursue the project with the goal of building Asia's largest AI infrastructure based on close coordination with Nvidia.

An AI factory is an 'intelligence factory' that continuously produces 'tokens'—the core unit of AI—using power and data as raw materials, and it is constructed based on the Nvidia DSX-powered infrastructure. This is a next-generation concept that goes beyond existing data centers limited to general-purpose computing and data storage.

The AI factory is planned to begin operations in South Korea in 2027. This is expected to become the first case to verify the governance and operational structure of the two companies' AI cloud. SKT plans to gradually scale this model into GW-grade infrastructure and expand the AI infrastructure across Asia.

The cooperation with Nvidia is projected to serve as a core foundation for SKT's AI cloud business. Starting with Nvidia Blackwell GPUs, SKT plans to support AI training and inference, and will sequentially utilize the latest Nvidia Vera Rubin platform, which is scheduled to be supplied in the second half of this year.

Implementing an AI factory requires 'AI full-stack' capabilities that encompass not only computing but also networking and software. Through this partnership, SKT will secure Nvidia's AI infrastructure, including computing and dedicated software, while Nvidia will be able to utilize SK Hynix's memory leadership alongside SKT's expertise in AI factory construction and operation.

SKT intends to rapidly grow its AI cloud business using this partnership as a springboard. Unlike the existing cloud business that provides general-purpose cloud solutions, AI cloud is a new business model that delivers services specialized for AI workloads, such as AI training, inference, and data processing.

This market has recently emerged as a key pillar of the global AI infrastructure supply chain, to the extent that global Big Tech companies are leasing AI infrastructure on a massive scale.

SKT's vision is to leap forward as one of Asia's leading AI cloud providers by establishing itself beyond South Korea as a core partner that substantially provides the AI infrastructure and business networks required by Nvidia throughout Asia.

Furthermore, SK Group and Nvidia announced plans to pursue research to jointly develop next-generation AI factory architectures, moving beyond mere infrastructure construction.

The existing cooperation between SK and Nvidia had focused on the semiconductor sector, such as SK Hynix's High Bandwidth Memory (HBM), but through this agreement, the cooperative relationship between the two companies expands across the entire spectrum of AI infrastructure, including AI factory construction and operation.

The new R&D cooperation includes joint research on a new computing architecture that enhances the performance of both GPUs and memory simultaneously from the design stage. To achieve this, the two companies plan to form a joint committee.

SK Group Chairman Chey Tae-won stated, "Based on our close partnership with Nvidia, we have secured competitiveness in full-stack AI infrastructure that spans from chips to data center operations," adding, "Moving beyond simply providing services, both companies will jointly respond to GPU, memory, and energy issues, thereby reborn as a representative AI cloud provider leading the development of the AI ecosystem across Asia."

Nvidia CEO Jensen Huang remarked, "Telecommunication networks are evolving into national AI infrastructure," adding, "The telecom networks connecting people, businesses, and devices are now becoming the foundation of the AI cloud. Through the Nvidia DSX platform, SKT will be able to build large-scale AI clouds and deliver agent AI, enterprise AI, and physical AI to leading enterprises and industries in Korea and worldwide."

                                                                                                             Ahn So-yeon