Chip giant Nvidia and South Korean conglomerate SK Group announced on Saturday, 25 July 2026 a joint initiative valued at over $500 billion, covering next-generation AI factories and advanced memory development, with South Korea positioned as a central node in global AI infrastructure.

Context

SK Telecom, South Korea's largest mobile operator, will build a 2-gigawatt Nvidia Vera Rubin DSX AI Factory in Korea designed to serve global compute demand. The facility represents one of the largest announced AI data centre projects tied to a single chip architecture. Separately, SK Hynix — the world's leading producer of high-bandwidth memory — and Nvidia will jointly develop next-generation AI memory, including HBM (High-Bandwidth Memory), which is critical to powering modern AI training clusters.

Nvidia's post stated: 'Korea is ready to become a global AI powerhouse with world-class networks and data centers, leadership in chip technology and vast industrial scale.' The announcement signals a deepening of a supply-chain relationship that has grown substantially as AI workloads have scaled.

Policy Backdrop

South Korea has pursued AI and semiconductor leadership as a national priority since at least 2019, when its government launched the National AI Strategy to channel investment into chip fabrication, data infrastructure, and AI research. SK Hynix began volume production of HBM3 memory in 2022, supplying components integral to Nvidia's GPU-based AI accelerators used in data centres worldwide.

The latest initiative extends that trajectory, moving from a supplier relationship to a co-development partnership. This shift gives South Korea a more active role in defining the architecture of next-generation AI memory, not merely manufacturing to specification.

Stakeholders and Impact

For SK Group, the deal spans two of its most strategically important subsidiaries: SK Telecom on the infrastructure and connectivity side, and SK Hynix on the silicon side. Together, they position the conglomerate as a vertically integrated AI infrastructure partner for Nvidia rather than a component vendor. The 2-gigawatt scale of the proposed AI factory underscores the enormous energy and capital requirements of frontier AI compute.

For the broader semiconductor and AI industry, the announcement reflects a wider pattern of cross-border alliances aimed at addressing surging demand for AI compute capacity. Memory has emerged as a critical bottleneck in AI system performance, making the SK Hynix–Nvidia co-development agreement particularly significant for system architects and cloud providers globally.

What's Next

Attention will now turn to implementation timelines for the Vera Rubin DSX AI Factory, the energy and grid infrastructure required to support a 2-gigawatt facility, and the technical roadmap for next-generation HBM co-developed by SK Hynix and Nvidia. Regulatory approvals, land acquisition, and power procurement in South Korea will be among the near-term milestones to watch.

If executed at the announced scale, the initiative could meaningfully shift global AI compute capacity toward East Asia, complementing existing data centre clusters in the United States and Europe and reinforcing South Korea's ambition to be a tier-one AI nation.