NAVER, Brookfield and NVIDIA plan to scale up South Korea’s national AI “factory” capacity

NAVER, Brookfield and NVIDIA said they are expanding South Korea’s “national AI factory” infrastructure, a move aimed at increasing domestic capacity for AI compute. The announcement ties the expansion to NVIDIA’s DSX AI factory platform, which packages data center hardware and networking into factory-style deployments meant to accelerate the roll-out of large-scale AI systems.

In the plan described, the initial DSX deployment in Korea is set to grow from 55 megawatts of power capacity to 200 megawatts. Megawatts measure the electrical draw a data center uses, and scaling from 55 MW to 200 MW indicates a much larger footprint for training and running advanced AI workloads.

The partners did not lay out, in the information available from the announcement coverage, a detailed timeline for when the additional capacity would be reached, nor did they specify a phase-by-phase delivery schedule for the incremental power. They also did not provide a breakdown of the expected mix of AI training versus inference workloads or how capacity would be allocated across customer or government use cases.

NVIDIA’s role centers on DSX, a deployment approach built around standardized AI infrastructure. DSX is described as an “AI factory” model rather than a traditional one-off data center build, with the goal of reducing time and complexity for customers that want to deploy large-scale AI quickly. For NVIDIA, such factory deployments are strategically relevant because they convert demand for chips and systems into longer-lived infrastructure programs tied to data center capacity.

Brookfield is positioned in the announcement as a partner for expanding the physical infrastructure needed to host the compute. Brookfield’s involvement aligns with its broader investment and development profile in energy-intensive facilities such as data centers, where scaling power, cooling and physical builds are often as significant as the compute hardware itself.

For NAVER, the expansion underscores continued emphasis on domestic AI capacity. As a major South Korean technology and internet company, NAVER has been active in building AI services and platforms, and access to higher-capacity compute can reduce bottlenecks for model development and AI applications that require substantial training and ongoing compute.

What remains unclear is how the expanded capacity will be governed. The public coverage did not specify what portion of the new megawatts would be reserved for NAVER versus shared tenants, whether the arrangement is structured as sovereign-only capacity, or what service terms would apply. It also did not disclose any spending figures, contract value, or the specific technical configuration beyond the DSX factory expansion and the power-capacity target.

Looking ahead, the key details to watch are the announced delivery schedule for reaching the 200 MW level, the degree of public versus private access to the capacity, and whether the partners provide additional performance or capacity milestones. Those elements will determine how quickly the buildout translates into measurable AI capability for users inside South Korea.