Nvidia and SK Group’s $500B AI partnership builds massive memory factories

A $500 Billion Blueprint: Nvidia and SK Group Forge Future of AI Infrastructure

A massive collaboration is underway between Nvidia and SK Group, cementing a strategic relationship valued at more than $500 billion. This partnership is not just about semiconductors; it is a blueprint for building the next generation of global artificial intelligence infrastructure.

At the heart of this monumental deal is a long-term memory supply agreement between Nvidia and SK hynix, ensuring the flow of essential memory components. This foundational agreement supports SK Telecom’s ambitious vision to establish a colossal 2-gigawatt AI data center in South Korea, setting the stage for future expansions of advanced computing systems across the Asia-Pacific region.

The ambition driving this project is significant: the new facility will leverage Nvidia’s cutting-edge DSX AI factory platform alongside Vera Rubin accelerated computing systems, powered by SK hynix HBM4 memory. This infrastructure is designed to facilitate diverse applications, ranging from sovereign AI and enterprise solutions to physical and agentic AI deployments.

The scale of this collaboration extends far beyond hardware supply and deployment. The $0.5 trillion aggregate value reflects the expected commercial activity spanning years—from SK Telecom’s procurement of Nvidia GPUs and networking equipment to the revenue generated by ecosystem partners involved in building out the essential AI factory infrastructure, including specialized manufacturing, cooling, and power solutions.

Inside the planned 2-gigawatt data center, the sheer computational density is staggering. The facility is envisioned to house thousands of NVL72 VR200 racks and hundreds of thousands of Vera CPUs and Rubin AI GPUs. This infrastructure aims not only for massive processing power but also for peak energy efficiency, utilizing Nvidia’s DSX platform which is engineered as a complete factory blueprint combining hardware, software, networking, and partner technologies.

Nvidia’s DSX framework is designed to deliver the lowest-cost token generation while maximizing energy efficiency in data center operations. These systems are engineered to handle immense power requirements; for instance, an NVL72 VR200 rack can operate with power consumption ratings of up to 166 kW or 240 kW.

While the exact configuration and Power Usage Effectiveness (PUE) of the specific SK Telecom facility remain under development, the intention is clear: establishing a regional hub where next-generation AI will be developed and deployed at an unprecedented scale. This strategic alignment positions the companies to lead the global shift towards AI-driven infrastructure.