NVIDIA, SK Telecom Launch Gigawatt AI Cloud for Korea

NVIDIA and SK Telecom are joining forces to construct a substantial AI cloud infrastructure in South Korea, aiming for a gigawatt-scale deployment. This ambitious project, slated for its initial phase to launch by 2027, is designed to bolster the nation's artificial intelligence capabilities, supporting everything from sovereign AI initiatives to advanced agentic applications for a wide array of industries. The collaboration leverages SK Telecom's extensive telecommunications network and data center expertise alongside NVIDIA's cutting-edge AI hardware and software stack.

The core of this initiative revolves around the concept of 'AI factories,' specialized facilities engineered for the efficient production of AI intelligence, often referred to as tokens. Unlike conventional cloud services, these AI factories are meticulously optimized for the demanding computational requirements of AI model training and inference, a critical distinction in the rapidly evolving AI sector. The platform will be built upon NVIDIA's DSX full-stack reference architecture, a comprehensive system integrating hardware, software, and operational protocols to enhance energy efficiency and reduce the cost of generating AI outputs.

This strategic alliance underscores South Korea's prominent position in AI adoption and technological advancement. The country is a global hub for innovation across sectors such as semiconductors, telecommunications, manufacturing, and mobility, all of which are increasingly integrating AI into their operational frameworks. The planned AI cloud is expected to accelerate this integration, providing the necessary infrastructure for Korean enterprises to develop and deploy sophisticated AI solutions.

The partnership extends beyond infrastructure. SK Telecom is set to become an official NVIDIA Cloud Partner, integrating into a global network that utilizes NVIDIA's latest AI technologies. Furthermore, both companies intend to deepen their collaboration into research and development, focusing on next-generation AI factory architectures. This includes exploring advancements in accelerated computing, memory technologies, and data center management to create more scalable, resilient, and efficient AI services for the Asian market and beyond.

This development aligns with SK Telecom's ongoing commitment to AI innovation. The company has previously demonstrated its work with digital twins for semiconductor facilities using NVIDIA Omniverse and has adopted open-source NVIDIA datasets for its own AI model development, notably for the Korean government's Sovereign AI Foundation Model Project. The scale of this new AI cloud project signifies a significant step-up in their commitment to shaping the future of AI infrastructure.

The initiative addresses a critical need for specialized, high-capacity AI computing resources. As AI models grow in complexity and data volumes expand, the demand for GPU-accelerated infrastructure is projected to surge. Industry analysts predict a compound annual growth rate exceeding 30% for the global AI infrastructure market over the next five years, highlighting the strategic importance of such large-scale deployments. This collaboration positions both NVIDIA and SK Telecom to capture a significant share of this expanding market.