The $500 Billion Blueprint: NVIDIA and SK Group Redefine AI Infrastructure
In a move that signals a new phase of industrial-scale AI deployment, NVIDIA has entered into a strategic partnership with South Korea's SK Group, with a combined investment value exceeding $500 billion. The alliance focuses on co-developing the foundational hardware and facilities required for the next generation of artificial intelligence.
Two Pillars of the Partnership
The collaboration is built on two interconnected fronts. The first is the construction of massive AI data center capacity. SK Telecom, under the SK Group umbrella, will lead the development of an AI data center with a total power capacity of 2 Gigawatts (GW). This facility is designed from the ground up to be powered by NVIDIA's upcoming Vera Rubin AI platform.
The second, equally critical pillar is memory. NVIDIA is establishing a long-term supply and co-development agreement with SK Hynix, SK Group's semiconductor arm. The objective is to secure a reliable pipeline for next-generation High Bandwidth Memory (HBM) and to jointly optimize HBM specifications specifically for demanding workloads like AI training, AI agents, and physics-based AI simulations.
The HBM4 Timeline Comes into Focus
A key technical revelation from the announcement is the confirmed adoption timeline for HBM4, the forthcoming standard in high-bandwidth memory. The planned 2GW data center will integrate both NVIDIA's Vera Rubin chips and SK Hynix's HBM4, marking one of the first major commercial projects to publicly commit to this advanced memory technology.
HBM4 is widely anticipated within the industry as the solution to the escalating memory bandwidth demands of future AI models. Its deployment is expected to significantly alleviate data transfer bottlenecks between processors and memory, a critical hurdle known as the "memory wall."
Looking Ahead to 2027
The partners have set a clear milestone: the first phase of this next-generation AI data center is scheduled to come online in 2027. This date provides a tangible horizon for the industry to anticipate the materialization of these advanced hardware synergies.
The scale of this investment, surpassing half a trillion dollars, underscores a strategic shift. It represents a deep, vertical integration between leading AI compute and memory architects, potentially setting a new standard for how AI infrastructure is built at a global scale.