www.industry-asia-pacific.com
27
'26
Written on Modified on
NAVER, NVIDIA, and Brookfield Expand Sovereign AI Infrastructure in Korea
NAVER, NVIDIA, and Brookfield collaborate to expand South Korea's sovereign AI factory infrastructure to 200 megawatts at the GAK Sejong data center.
www.nvidia.com
NAVER, NVIDIA, and Brookfield have announced a proposed expansion of South Korea's sovereign AI factory infrastructure, scaling the initial buildout at NAVER’s GAK Sejong hyperscale data center from 55 megawatts to 200 megawatts. The initiative utilizes the NVIDIA DSX platform, incorporating NVIDIA Vera Rubin and Blackwell computing architectures to deliver multi-tenant AI cloud capacity for enterprises, startups, and government organizations.
Context of the Cooperation
The rapid acceleration of artificial intelligence adoption across the global economy requires access to trusted, sovereign, and scalable computing infrastructure. Developing national AI models, autonomous agents, and digital twins presents immense computational, capital, and data center operational demands that single entities cannot easily scale independently.
To address these technical and financial requirements, cooperation was established between NAVER Corp., NVIDIA, and Brookfield. The partnership combines NAVER's full-stack AI and hyperscale data center operating expertise with NVIDIA's accelerated computing platform and Brookfield's global AI infrastructure investment capabilities. Haejin Lee, founder and chairman of NAVER, stated that strategic investments and infrastructure agreements propel NAVER's AI factory vision into execution to strengthen South Korea's AI competitiveness. Sikander Rashid, global head of AI infrastructure at Brookfield, noted that sovereign and scalable AI infrastructure is a critical strategic priority. Jensen Huang, founder and CEO of NVIDIA, added that AI factories represent essential national infrastructure to compete and innovate in the intelligence era.
Technical Solution and Responsibilities
The technical framework centers on expanding the NVIDIA DSX platform—an end-to-end, codesigned architecture purpose-built to minimize token costs and accelerate deployment. Responsibilities are divided across the participating partners to align computing hardware, capital financing, and facility operations:
- NAVER operates the GAK Sejong data center, manages multi-tenant AI cloud operations, and funds remaining capital requirements for the project.
- NVIDIA supplies accelerated hardware platforms—including NVIDIA Vera Rubin and NVIDIA Blackwell silicon—and plans a $1 billion investment into NAVER Corp., subject to customary closing conditions and financing thresholds.
Brookfield provides capital financing under a nonbinding term sheet to fund up to $9 billion, leveraging its AI infrastructure investment platform.
At a system level, the NVIDIA DSX platform integrates hardware chips, software stacks, facilities, and networking frameworks. The software layer incorporates NVIDIA DSX MaxLPS™ to maximize token throughput per megawatt, alongside NVIDIA DSX OS™ for lifecycle management, health automation, system resiliency, and multi-tenant AI factory oversight. Furthermore, NAVER is adapting its HyperCLOVA X models to be based on NVIDIA Nemotron™ 3 Ultra open models, while developing a Seoul World Model using urban spatial data on NVIDIA Cosmos™ world foundation models.
Deployment or Implementation
The deployment strategy centers on expanding the NVIDIA DSX AI factory buildout at NAVER’s GAK Sejong hyperscale data center in Sejong, South Korea. The project expands the initial 55-megawatt buildout to 200 megawatts, establishing a long-term operational roadmap toward gigawatt-scale sovereign AI infrastructure.
During the implementation phase, NAVER will deploy NVIDIA Agent Toolkit software, including NVIDIA NemoClaw™ blueprints, to launch an AI agent platform in South Korea. The expanded infrastructure will supply production-scale AI compute to innovators across South Korea and the United States, supporting the pretraining, post-training, and reinforcement learning of next-generation AI models, autonomous agents, and industrial services.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
At a system level, the NVIDIA DSX platform integrates hardware chips, software stacks, facilities, and networking frameworks. The software layer incorporates NVIDIA DSX MaxLPS™ to maximize token throughput per megawatt, alongside NVIDIA DSX OS™ for lifecycle management, health automation, system resiliency, and multi-tenant AI factory oversight. Furthermore, NAVER is adapting its HyperCLOVA X models to be based on NVIDIA Nemotron™ 3 Ultra open models, while developing a Seoul World Model using urban spatial data on NVIDIA Cosmos™ world foundation models.
Deployment or Implementation
The deployment strategy centers on expanding the NVIDIA DSX AI factory buildout at NAVER’s GAK Sejong hyperscale data center in Sejong, South Korea. The project expands the initial 55-megawatt buildout to 200 megawatts, establishing a long-term operational roadmap toward gigawatt-scale sovereign AI infrastructure.
During the implementation phase, NAVER will deploy NVIDIA Agent Toolkit software, including NVIDIA NemoClaw™ blueprints, to launch an AI agent platform in South Korea. The expanded infrastructure will supply production-scale AI compute to innovators across South Korea and the United States, supporting the pretraining, post-training, and reinforcement learning of next-generation AI models, autonomous agents, and industrial services.
Edited by Romila DSilva, Induportals Editor, with AI assistance.
