NVIDIA and SK Group Plan $500 Billion AI Infrastructure Expansion

The SK Group and NVIDIA partnership will invest more than $500 billion in AI infrastructure, accelerated computing, and next-generation memory technology.

The companies signed letters of intent covering a large AI cloud project in South Korea and a long-term memory supply agreement between NVIDIA and SK hynix.

SK Telecom plans a 2-gigawatt AI cloud

SK Telecom plans to build a 2-gigawatt AI cloud that will provide large-scale computing resources to customers across South Korea and the Asia-Pacific region.

The infrastructure will use the NVIDIA DSX platform and Vera Rubin accelerated computing systems. The first AI factory should begin operations in 2027.

The project will support several types of artificial intelligence services, including sovereign AI, enterprise AI, agentic AI, and physical AI.

SK Telecom plans to make the infrastructure available to more organizations that need substantial computing capacity for training and running AI models.

NVIDIA DSX will power the AI factory

The AI factory will use NVIDIA’s full-stack DSX architecture, which combines accelerated computing hardware, systems, software, and technologies from NVIDIA partners.

NVIDIA designed the platform to lower the cost of generating AI tokens while improving power efficiency. These improvements could help companies operate larger AI workloads without increasing energy consumption at the same rate.

The Vera Rubin systems will provide the accelerated computing foundation for the project. NVIDIA developed the platform for demanding AI workloads, including large language model training, autonomous agents, and physical AI applications.

SK hynix will supply HBM4 memory

SK hynix will provide HBM4 memory for the Vera Rubin systems deployed through the partnership.

The memory manufacturer will also establish a long-term AI memory partnership with NVIDIA. The agreement aims to give NVIDIA a more stable supply of next-generation high-bandwidth memory as demand for AI computing continues to increase.

HBM plays a critical role in AI accelerators because it allows processors to access large amounts of data quickly. More advanced memory can help reduce bottlenecks when training and running complex AI models.

Companies will develop future AI memory

NVIDIA and SK hynix also plan to jointly develop and optimize future memory technologies for artificial intelligence.

The collaboration will focus on high-bandwidth memory designed for large AI workloads. These technologies will support large language models, agentic AI systems, and physical AI applications that interact with real-world environments.

Working together could allow NVIDIA to optimize its computing platforms around upcoming SK hynix memory products. SK hynix could also design future memory around the requirements of NVIDIA’s next-generation accelerators.

South Korea will become a major AI infrastructure center

The partnership aims to meet growing demand for AI computing throughout the Asia-Pacific region, with South Korea serving as a major infrastructure center.

SK Group wants the investment to strengthen South Korea’s position as a global hub for AI infrastructure and innovation.

NVIDIA believes the country’s telecommunications networks, data centers, semiconductor expertise, and industrial capacity provide a strong foundation for further AI development.

The partnership combines SK Telecom’s network and cloud infrastructure, SK hynix’s memory technology, and NVIDIA’s accelerated computing platforms.

In other news, the NVIDIA Vera CPU is already helping the company design and verify its next-generation CPUs and GPUs.