NVIDIA and SK Group Plan $500 Billion AI Initiative
Published July 30, 2026
NVIDIA and South Korea's SK Group are significantly expanding their cooperation on AI infrastructure and high-performance memory. The companies describe an initiative worth more than $500 billion, ranging from new AI factories to joint development of future AI memory.
A central component is a new AI infrastructure planned by SK Telecom with a capacity of up to two gigawatts. NVIDIA's DSX platform and the upcoming Vera Rubin generation are slated for deployment.
The first AI factory is scheduled to go online in 2027. Hardware will include memory from SK hynix, notably HBM4.
NVIDIA secures access to critical AI memory
Beyond compute infrastructure, the partnership with SK hynix plays a key role. NVIDIA and the memory maker intend to deepen their long-term collaboration and jointly develop and optimize memory solutions for upcoming AI systems.
A particular focus is High Bandwidth Memory (HBM). This memory type is critical for modern AI accelerators because training and running large models requires moving enormous data volumes between memory and compute units as fast as possible.
The agreement is also meant to give NVIDIA more stable, long-term access to future memory technologies.
South Korea as an AI infrastructure hub
With this project, NVIDIA and SK are targeting not only the Korean market. The planned infrastructure is designed to serve growing demand for AI compute across the broader Asia-Pacific region.
Supported applications include agentic AI, physical AI, enterprise AI and sovereign AI infrastructure.
For SK Group, the partnership is unusually broad in scope: SK Telecom provides data centers and cloud infrastructure, while SK hynix supplies one of the most important building blocks of modern AI hardware — high-performance memory.
If the announced plans are implemented at this scale, South Korea could significantly strengthen its position in the global AI infrastructure landscape.
