NAVER (Technology company) and NVIDIA (Technology Company) have announced a major expansion of South Korea's sovereign AI infrastructure, with NAVER committing to build AI factories at gigawatt scale using the NVIDIA DSX platform, starting with an initial 55-megawatt deployment at its GAK Sejong data center in Sejong, South Korea.

The partnership, announced June 7, 2026, positions NAVER to serve enterprises, industries and government organizations in Korea and globally with what the companies describe as full-stack, end-to-end AI production infrastructure.

The agreement also deepens an existing collaboration between the two companies on AI model development, including NAVER's HyperCLOVA X large language models.

GAK Sejong as the Foundation for Sovereign AI Scale

The initial phase of the expansion centers on NAVER's GAK Sejong facility, which the company describes as a next-generation, hyperscale data center designed for high-density NVIDIA accelerated computing. The facility is built with advanced automation, sustainability features, resilient energy-efficient operations, and robust disaster-response capabilities.

Starting at 55 megawatts, NAVER has stated plans to scale toward gigawatt-level capacity using the NVIDIA DSX platform. The platform is described as a purpose-built, end-to-end codesigned stack for AI factories, spanning chips, systems, software, facilities and partner technologies, with the stated goal of minimizing token cost and accelerating time to first production.

NVIDIA DSX MaxLPS software is designed to maximize token throughput per megawatt, while NVIDIA DSX OS provides an operating layer for lifecycle management, health automation, resiliency and multi-tenant AI factory management.

NAVER's existing expertise in building and operating large-scale proprietary GPU clusters and hyperscale data centers is expected to complement these platform capabilities.

Serving Korea's Industries and Global AI Cloud Customers

The expanded infrastructure is intended to serve a broad range of customers, including Korean enterprises, manufacturers, and government organizations, as well as global AI cloud clients.

NAVER already operates sovereign AI infrastructure in Europe and the Middle East and aims to offer government and enterprise customers a trusted alternative for secure, high-performance digital services that comply with local regulatory and data-sovereignty requirements.

Jensen Huang, founder and CEO of NVIDIA, said useful AI has arrived and that demand for AI factories is extraordinary. Huang stated that NVIDIA DSX would help Korea scale sovereign intelligence infrastructure for what he described as the agentic era, covering AI agents, AI factories, and physical AI.

Haejin Lee, founder and chairman of NAVER, said the company is building sovereign AI infrastructure capable of serving Korea's industries and global customers with trusted, high-performance AI.

Lee stated that the NVIDIA DSX platform would help customers move from AI experimentation to production-scale AI factories powering models, agents and real-world services.

HyperCLOVA X and the NVIDIA Nemotron Coalition

The infrastructure expansion builds on an existing model development collaboration between NAVER and NVIDIA. NAVER is advancing its HyperCLOVA X models by fine-tuning the NVIDIA Nemotron 3 Ultra open model using its proprietary data and training expertise.

The companies say this process is delivering a more capable and culturally fluent model for Korean and global enterprise customers. HyperCLOVA X models based on NVIDIA Nemotron are also described as a platform to support sovereign AI initiatives in Europe and the Middle East.

NAVER has been identified as the first Korean company to participate in the NVIDIA Nemotron Coalition, an initiative focused on contributing to open model development across pretraining, post-training and reinforcement learning. As part of this engagement, NAVER plans to launch an AI Agent Platform in Korea in the second half of 2026, powered by NVIDIA NemoClaw blueprints.

NAVER holds a notable position in the global AI landscape, having been identified as the world's third company to develop a hyperscale large language model. The company has been providing internet services and IT infrastructure since 1999 and has grown to offer end-to-end capabilities across the AI value chain, including AI services, foundational AI infrastructure, supercomputing resources and cloud platforms through its public cloud service, NAVER Cloud Platform.

Seoul World Model and Physical AI Development

Beyond language models, NAVER is also developing what it calls a Seoul World Model, which uses the company's proprietary urban street-view data and spatial modeling technology. This project is being built on NVIDIA Cosmos world foundation models and represents NAVER's entry into physical AI development.

The Seoul World Model initiative reflects a broader industry trend toward AI systems capable of understanding and operating within real-world physical environments. NAVER's access to extensive urban spatial data from its mapping and street-view operations provides the proprietary data foundation for this effort.

AI Factories as Critical Infrastructure

Both companies framed the announcement within a broader shift in how the AI industry is evolving. As AI moves increasingly into production use cases, AI factories are being positioned as critical infrastructure for training, post-training and inference workloads, rather than experimental computing resources.

The NVIDIA DSX platform is described as specifically designed to support this transition, providing what the companies say is a proven AI factory blueprint for providers seeking to build and operate at scale while managing token costs and operational complexity.

South Korea Data Center Pipelines: Stay Ahead of Every Shovel in the Ground

The South Korea data center boom is accelerating faster than most market participants can track; hyperscaler expansions, colocation campuses, and edge deployments are reshaping the competitive landscape across dozens of markets simultaneously. If your intelligence is lagging behind the activity, you are already behind the deal.

The Global Project Tracking (GPT) platform by Blackridge Research was built precisely for this environment, aggregating project-level intelligence across South Korea into a single, structured, continuously updated resource. Whether you are identifying land acquisition targets, benchmarking contractor activity, or mapping supply chain opportunities, the platform delivers the granular data your team needs to act with confidence.

From early planning signals through to completed facilities, the GPT platform gives you end-to-end visibility across the full South Korea data center development cycle, so your business development, investment, and procurement teams are always working from the most current picture available.

• Upcoming Projects

• Tender Notices

• Contract Awards

• Projects Under Construction

• Completed Projects

To explore how the Global Project Tracking (GPT) platform can sharpen your South Korea data center market intelligence, contact the Blackridge Research team today and Book a Free Demo.