NVIDIA and SK Group Broaden Alliance to Build Mega-Scale AI Factories and Co-Develop Next-Gen Memory

SK Group and Nvidia unveiled a partnership exceeding $500 billion to expand AI infrastructure, centered on a 2-gigawatt “AI factory” in South Korea slated to begin operations in 2027. SK Telecom will deploy Nvidia’s Vera Rubin accelerated computing and DSX platform, while SK hynix and Nvidia struck a long-term pact to secure and co-develop next-generation high-bandwidth memory, including HBM4. The initiative aims to meet surging global compute demand and fuel sovereign, enterprise and agentic AI across the Asia-Pacific region, with an emphasis on lowering token costs and boosting energy efficiency. The companies cast the plan as positioning Korea as a global AI hub. Letters of intent outline the framework, and both sides flagged typical forward-looking risks, including supply-chain dependencies, technology shifts and market acceptance. The move underscores intensifying competition to lock down compute capacity and memory supply as AI model training scales.

Read more