An AI cloud refers to large-scale AI infrastructure made up of AI factories that “manufacture tokens”—the basic building blocks of intelligence—by processing data at scale.

Unlike conventional cloud providers that offer general-purpose computing services, AI clouds are purpose-built for GPU-accelerated workloads across training, inference, and agentic AI applications.

SK Telecom’s AI cloud will be built on NVIDIA’s DSX full-stack reference architecture, integrating software, hardware, and operational systems designed to deliver the lowest token costs with maximum energy efficiency.

The AI cloud will support training, inference, and agentic workloads, including sovereign AI, physical AI, and enterprise AI services for industries across South Korea, with plans to expand further into Asia.

The initiative underscores South Korea’s position as one of the world’s leading pro-AI industrial economies. Korean companies are at the forefront of telecommunications, semiconductors, memory, manufacturing, robotics, mobility, and consumer technology—sectors where AI is rapidly moving from research environments into secure, production-scale deployments.

“Telecom networks are becoming national AI infrastructure,” said Jensen Huang, founder and CEO of NVIDIA. “They connect people, companies, devices and machines—and now they can become the backbone of new AI clouds. With NVIDIA DSX, SK Telecom can build Korea’s AI cloud at scale and bring agents, enterprise and physical AI to the companies and industries that power Korea and the world.”

“Through our close partnership with NVIDIA, we have now secured full-stack AI infrastructure capabilities, from chips to data center operations,” said Chey Tae-won, Chairman of SK Group. “We will work with NVIDIA to address GPU, memory, and energy challenges and become a leading AI factory player shaping Asia’s AI ecosystem.”

SK Telecom AI Cloud to Power Agentic and Physical AI

The new AI cloud is designed to meet rising global demand for advanced AI infrastructure, as accelerated computing needs continue to grow rapidly.

The initiative builds on SK Telecom’s work in agentic AI and physical AI development. At GTC Taipei, the company showcased its use of digital twins for SK hynix semiconductor fabs using NVIDIA Omniverse libraries, aimed at optimizing complex, large-scale manufacturing environments.

In April, SK Telecom also announced its adoption of NVIDIA Nemotron™ open-source datasets to train its A.X K1 model under the Korea government’s Sovereign AI Foundation Model Project.

As part of this collaboration, SK Telecom will join the NVIDIA Cloud Partner program, gaining access to NVIDIA’s latest AI infrastructure, software, and developer ecosystem to deliver high-performance and cost-efficient AI cloud services globally.

NVIDIA DSX Platform and Next-Generation AI Infrastructure

The NVIDIA DSX platform is designed for end-to-end AI factory optimization, integrating chips, systems, software, facilities, and partner technologies to minimize token costs and accelerate deployment.

Its DSX MaxLPS software aims to optimize token output per megawatt, reducing costs while improving efficiency. Meanwhile, NVIDIA DSX OS™ provides an open, modular operating layer for lifecycle management, runtime consistency, automation, resiliency, and multi-tenant AI operations.

NVIDIA and SK Group also announced plans for joint research into next-generation AI factory architecture, focusing on silicon-to-grid innovation, including advancements in computing, memory technologies, and data center operations. The collaboration aims to further optimize large-scale AI infrastructure for greater efficiency, scalability, and resilience.