Key Takeaways
- Sector: Artificial Intelligence (AI), Digital Infrastructure.
- Geography: South Korea, Global.
Analysis
South Korean tech giant NAVER is significantly bolstering its artificial intelligence capabilities by partnering with NVIDIA to construct expansive AI data processing facilities. This strategic alliance will initially see a substantial 55-megawatt deployment at NAVER's GAK Sejong data center, with ambitious plans to scale this infrastructure to gigawatt levels. The move is a direct response to the escalating global demand for advanced AI services, including sovereign AI initiatives, enterprise-grade AI workloads, and government-specific applications.
The core objective of this collaboration is to empower organizations to transition from AI experimentation phases to full-scale production deployments. NAVER aims to deliver trusted, high-performance AI services that adhere to stringent regional data sovereignty regulations. This initiative is particularly crucial for nations seeking to establish independent AI infrastructure, enabling them to control their data and AI development.
NVIDIA's CEO, Jensen Huang, underscored the extraordinary demand for AI factories, emphasizing that NAVER's new facilities will be instrumental in supporting developers, businesses, and various industries within Korea and beyond. He highlighted the NVIDIA DSX platform's role in enabling countries to build sovereign intelligence frameworks, supporting everything from AI agents to complex physical AI systems.
This expansion is not merely about hardware; NAVER intends to leverage NVIDIA's comprehensive AI technology stack to advance its own strategic projects. This includes enhancing its proprietary HyperCLOVA X large language models, developing next-generation agentic AI services, and creating a sophisticated Seoul World Model. The latter will utilize proprietary urban street-view data and advanced spatial modeling techniques, showcasing a commitment to localized AI innovation.
The partnership deepens an existing relationship focused on sovereign and physical AI. NAVER is actively refining its HyperCLOVA X models by fine-tuning NVIDIA's open Nemotron 3 Ultra model with its unique datasets and training expertise. This effort is designed to yield superior Korean-language processing capabilities while also supporting sovereign AI deployments in international markets, including Europe and the Middle East. NAVER has also joined the NVIDIA Nemotron Coalition, contributing to open model development across various AI training stages.
The NVIDIA DSX platform itself is engineered as a unified AI factory architecture, integrating hardware, software, and partner technologies. Its design prioritizes reducing operational costs, specifically token costs, and accelerating deployment timelines. NAVER will combine this platform with its extensive experience managing large-scale GPU clusters and hyperscale data centers. The deployment will incorporate NVIDIA DSX MaxLPS software for optimizing energy efficiency and throughput, alongside NVIDIA DSX OS software for robust lifecycle management, automation, and multi-tenant operations.