NAVER scales AI infrastructure to 55 MW, eyes gigawatt future

Lead

On June 7, 2026, NAVER disclosed that it will launch a sovereign AI infrastructure starting at 55 megawatts, with a roadmap to reach gigawatt‑scale capacity using NVIDIA’s DSX™ platform.

Context

The announcement comes as enterprises, governments and developers worldwide scramble for compute power to train and run ever larger models. NVIDIA’s newsroom notes that the DSX platform enables rapid design, construction and scaling of full‑stack AI systems, a capability NAVER intends to harness for a broad customer base that includes industry, public sector and corporate users.

South Korea is not alone in betting on sovereign AI. A week later, NVIDIA highlighted the United Kingdom’s own push, where the government and local partners are converting policy commitments into tangible infrastructure projects powered by NVIDIA technology. Both cases illustrate a growing belief that national‑level AI compute resources are essential for competitiveness and security.

Impact

Starting at 55 MW, NAVER’s rollout already rivals the power draw of a small city. Scaling to gigawatt levels would multiply that demand by an order of magnitude, implying massive electricity costs and a need for efficient cooling and power‑management solutions. While the sources do not disclose specific pricing, the sheer scale suggests that operating expenses will become a critical factor for NAVER and its customers.

By leveraging NVIDIA DSX, NAVER aims to offset some of those costs through hardware‑level optimizations and integrated software stacks that streamline model training and inference. The platform promises end‑to‑end support, which could reduce the overhead of stitching together disparate tools, thereby lowering operational complexity for enterprises that adopt the service.

For the broader market, NAVER’s move signals that non‑U.S. players are building home‑grown AI compute ecosystems. This could diversify the supply chain, reduce reliance on foreign cloud providers, and potentially drive competitive pricing for AI workloads globally.

What’s Next

NAVER has not released a detailed timeline for the gigawatt expansion, but the partnership with NVIDIA suggests a phased approach that will likely involve incremental capacity upgrades as demand materializes. The company will also need to navigate regulatory and environmental reviews given the scale of power consumption.

Industry observers will watch how NAVER’s infrastructure competes with existing cloud giants and whether other regions will emulate the sovereign model. As more governments prioritize AI self‑sufficiency, the combination of massive power draws and sophisticated platforms like DSX may become a standard blueprint for future national AI strategies.