en.Wedoany.com Reported - On June 8, South Korean internet and cloud service company NAVER and US-based NVIDIA announced an expanded collaboration on AI infrastructure. NAVER will expand its proprietary AI infrastructure based on the NVIDIA DSX platform, starting with a 55-megawatt scale at the Sejong GAK Data Center in South Korea, with plans to gradually scale up to gigawatt levels. This will provide computing power services to South Korean enterprises, manufacturing sectors, government agencies, and global AI cloud customers.

This partnership connects NAVER's hyperscale data center operational capabilities with NVIDIA's AI factory reference architecture. The GAK Sejong Data Center, originally built to support NAVER's AI, cloud services, and large-scale digital platform operations, will, following the introduction of DSX, further focus its infrastructure development on high-density accelerated computing, cluster scheduling, energy efficiency optimization, multi-tenant management, and AI factory lifecycle operations and maintenance. The DSX platform covers chips, systems, software, facilities, and partner technologies, aiming to reduce the cost per token for large-scale training, post-training, inference, and agent workloads, while shortening the cycle from AI infrastructure design to deployment. For NAVER, this system can serve South Korea's domestic sovereign AI needs and also support manufacturing, public sector, and cross-regional enterprise customers in deploying high-performance AI clouds within compliant environments. As AI applications transition from experimental phases to production-grade deployment, enterprise customers no longer require only single-point model calls but need stable model training capabilities, low-latency inference resources, data security boundaries, industry-specific model fine-tuning capabilities, and sustainably scalable cloud infrastructure. NAVER's accumulated expertise in search, maps, cloud, data centers, and local language models is being reorganized into AI factory service capabilities.

Collaboration at the model level is also progressing concurrently. NAVER will leverage NVIDIA's full-stack AI platform to continue developing the next-generation HyperCLOVA X model and will fine-tune it based on NVIDIA's open-source Nemotron 3 Ultra model.

This arrangement positions NAVER's AI cloud not merely as a computing power expansion project but also encompasses the development of regional models, agent services, and physical AI data capabilities. HyperCLOVA X has long been tailored for the Korean language context, South Korean enterprise services, and localized generative AI scenarios. By integrating Nemotron 3 Ultra, NAVER can enhance its model capabilities using its proprietary data, training experience, and linguistic-cultural adaptation expertise, offering AI services with superior local semantic understanding to enterprises in South Korea and globally. NAVER has also become the first South Korean company to join the NVIDIA Nemotron Alliance, participating in open model development stages such as pre-training, post-training, and reinforcement learning. It plans to launch an AI agent platform in South Korea in the second half of this year. Concurrently, NAVER is developing a "Seoul World Model" based on its proprietary city street view data and spatial modeling technology, combining it with the NVIDIA Cosmos world foundation model to provide underlying capabilities for physical AI, robotics, urban spatial understanding, and real-world service simulation. The integration of computing power, models, agents, and spatial data within a single collaboration framework indicates that South Korea's AI infrastructure is evolving from traditional cloud resource expansion towards systematic platform construction for sovereign AI and industrial AI.

The collaboration between NAVER and NVIDIA is also elevating the competition in South Korea's AI infrastructure to a higher level. A gigawatt-level AI factory implies that the project will subsequently face multiple constraints, including power supply, liquid cooling systems, network interconnects, storage throughput, cluster reliability, and operational costs. The competitive focus of AI clouds will simultaneously rest on computing power scale, model capabilities, and infrastructure efficiency. If the expansion of the GAK Sejong Data Center proceeds smoothly, NAVER is expected to combine its localized AI model capabilities, cloud service capabilities, and the NVIDIA AI factory ecosystem to form a sovereign AI cloud node serving enterprise customers in South Korea and overseas.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com