SK Group and Nvidia Announce $500 Billion AI Infrastructure Deal

Partnership targets Asia-Pacific AI demand

SK Group and Nvidia have formalized a partnership exceeding $500 billion in value to construct AI infrastructure addressing escalating global compute requirements. The agreement, documented through letters of intent, encompasses both the development of AI factories and the supply and co-development of advanced AI memory technologies.

According to details first reported by AI Watch, SK Telecom—the telecommunications arm of SK Group—will establish a 2-gigawatt AI cloud facility in South Korea. The installation will utilize Nvidia's DSX platform alongside Vera Rubin accelerated computing technology, integrated with SK hynix HBM4 memory. The inaugural facility is slated to begin operations in 2027.

Technical architecture and efficiency goals

The planned AI factories will deploy Nvidia's DSX full-stack architecture, which combines accelerated computing infrastructure, integrated systems, software components, and partner technologies. This design prioritizes energy efficiency and cost reduction for AI processing workloads as demand intensifies across Asia-Pacific markets, particularly within South Korea.

The collaboration aims to accelerate deployment of large-scale AI infrastructure supporting enterprise applications, agentic AI systems, physical AI implementations, and sovereign AI initiatives. SK hynix has separately established a long-term agreement with Nvidia covering supply and joint optimization of next-generation AI memory, including high-bandwidth memory (HBM) designed for large language model training and diverse AI workloads.

Why it matters

This partnership represents one of the largest announced investments in AI infrastructure globally and signals South Korea's ambition to transition from an AI technology adopter to a production hub. The 2027 timeline and specific technical specifications—particularly the 2GW power capacity—provide concrete benchmarks for tracking Asia-Pacific AI infrastructure development. The deal also underscores the strategic importance of memory technology in AI competitiveness, with SK hynix's HBM positioning as a critical component in next-generation systems.

Leadership perspectives

SK Group chairman Chey Tae-won framed the partnership as essential to South Korea's competitive positioning in the AI era. "Competitiveness depends not just on how effectively AI is utilised, but on how much intelligence we can produce," Chey stated. He emphasized that combining SK hynix's memory capabilities with SK Telecom's infrastructure expertise would help establish Korea as a global AI innovation center.

Nvidia founder and CEO Jensen Huang highlighted South Korea's existing advantages, including advanced networks, data center infrastructure, chip technology leadership, and industrial capacity. "Together with SK Telecom and SK hynix, we are building a new generation of AI factories that will power Korea's next wave of growth," Huang said.

The announcement builds upon an existing relationship between the companies and follows earlier plans to expand AI infrastructure. In related developments, Naver, Nvidia, and Brookfield have proposed additional expansion of South Korea's national AI factory infrastructure.

Details were first reported by AI Watch.