A $500 billion partnership and a 2-gigawatt AI data center plan landed onstage in San Francisco. Those two numbers make this more than a corporate deal: they point to a national-level compute commitment.
At an AI summit on July 24, 2026, NVIDIA CEO Jensen Huang discussed an initiative valued at over $500 billion with South Korean President Lee Jae Myung and SK Group leaders. Reporting indicates SK Telecom intends to build a 2-gigawatt AI data center using NVIDIA Vera Rubin accelerators and SK hynix HBM4 memory, with operation targeted around 2027. Coverage also describes collaboration among NVIDIA, SK hynix and Samsung Electronics on chip and memory design.
Put simply, this is about raw compute at scale. A 2-gigawatt facility implies a data center complex with unusually large power and cooling needs, built to run dense AI accelerators continuously. Vera Rubin refers to NVIDIA’s latest AI accelerator family, and HBM4 is high-bandwidth memory designed to move massive model data fast. Pairing those technologies inside a national initiative signals a push for co-designed hardware, local supply chains, and tightly integrated systems rather than a loose reseller relationship.
Why that matters beyond hardware: a plan of this scale can change where the world’s largest training and inference workloads run. If realized, it will attract software platforms, enterprise AI customers, and startups that prefer proximity to specialized hardware and memory. It also creates leverage for South Korean firms in chip and memory value chains, and it gives the government a tangible asset to market as part of an AI-hub strategy.
There are immediate practical questions. The reporting comes from summit remarks and press coverage, not from a detailed public roadmap. Timeline, capital structure, and regulatory approvals remain to be published. The 2027 target is rapid for a project of this size, so watch for permitting, grid upgrades, and supplier commitments.
For business leaders and policymakers, the takeaway is clear: national-scale compute commitments are now a strategic tool, not just corporate growth plays. Where megawatts concentrate, talent, services, and geopolitical attention follow.
If this initiative proceeds as sketched, it will test how fast infrastructure, energy planning, and semiconductor design can align to host the next generation of AI work at country scale.