en.Wedoany.com Reported - On June 8, South Korea's SK Hynix and US-based NVIDIA advanced their AI cloud infrastructure cooperation, planning to build a gigawatt-level AI cloud in South Korea based on the NVIDIA DSX platform. The first AI factory is expected to commence operations in 2027. This collaboration targets training, inference, and agent workloads, serving sovereign AI, physical AI, and enterprise AI applications, while reserving infrastructure pathways for future expansion into broader regions of Asia.
SK Hynix will also join the NVIDIA Cloud Partner Program, integrating telecom operators' network, data center, and cloud service capabilities into the NVIDIA AI factory ecosystem.
The core of this collaboration lies in advancing AI cloud from ordinary GPU clusters to the "AI factory" form. DSX is not a single hardware product but a full-stack reference architecture for the design, construction, and operation of AI factories, covering accelerated computing, networking, storage, facility simulation, power dispatch, cooling systems, and lifecycle operations. For SK Hynix, telecom networks originally connected users, enterprises, and edge nodes. By entering AI factory construction, its role will further extend to regional computing power supply, sovereign data hosting, enterprise AI service delivery, and high-density computing operations. Training large models requires continuous large-scale clusters, while inference and agent applications emphasize lower latency, higher throughput, and deployment capabilities closer to data sources. Telecom operators' accumulated expertise in backbone networks, edge data centers, customer access, and compliant operations precisely aligns with the new demands of AI cloud transitioning to production-level deployment.
The first AI factory is planned to commence operations in 2027, marking the first-phase window for both parties to validate AI cloud governance, resource scheduling, and commercial operation models.
SK Group and NVIDIA also plan to conduct joint research on next-generation AI factory architectures, focusing on full-stack innovation spanning chips, memory, computing platforms, data center operations, and grid coordination. SK Hynix's capabilities in high-bandwidth memory and next-generation storage technologies will form tighter system-level integration with NVIDIA's accelerated computing platforms. SK Hynix will assume the role of AI cloud infrastructure construction and operation, extending South Korea's AI infrastructure advantages from the chip supply chain to regional cloud service capabilities. Competition among AI factories no longer depends solely on GPU count; power acquisition, liquid cooling efficiency, network architecture, storage bandwidth, cluster scheduling, and cost per token all impact the actual output of computing centers. By expanding their collaboration scope to "from chip to grid," SK Hynix and NVIDIA demonstrate that large-scale AI infrastructure is becoming a systems engineering endeavor involving semiconductors, communications, power, and cloud computing. As demand from Asian enterprises and government agencies for sovereign AI, localized models, and industry agents rises, South Korea is poised to leverage such AI cloud projects to strengthen its regional computing node status and transform telecom infrastructure into a new type of industrial infrastructure for the AI era.
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
