Nvidia enters long-term tech partnerships with SK hynix and SK Telecom

Seoul, South Korea - NVIDIA Corp. announced separate long-term technical partnerships with SK Hynix Inc. and SK Telecom Co. on Monday to co-develop memory components and establish artificial intelligence data infrastructure.

The arrangements prioritize integrating Nvidia's computing platforms and hardware with the memory manufacturing and telecommunications infrastructure of the South Korean conglomerates to manage global supply requirements for high-performance computing.

Under the long-term technology partnership, SK Hynix will co-develop memory solutions for Nvidia's forthcoming hardware lines, including the Vera Rubin AI supercomputer, Vera CPUs, RTX Spark PCs, and the Jetson Thor robotics platform. 

The agreement seeks to align the development cycles of advanced memory components with Nvidia's long-term hardware roadmap.

SK Hynix is also embedding Nvidia's software into its semiconductor manufacturing and design workflows. 

The memory maker is deploying Nvidia CUDA-X libraries and the PhysicsNeMo framework to accelerate internal technology computer-aided design, computational lithography workflows, and engineering code simulations.

For its production operations, SK Hynix is building a factory digital twin to visualize and simulate its fabrication facilities. 

The system combines OpenUSD pipelines with Nvidia Omniverse libraries, utilizing the cuOpt optimization engine and Metropolis platform to coordinate autonomous mobile robots and automated factory assets.

Separately, Nvidia partnered with SK Telecom to construct a gigawatt-scale AI cloud network in South Korea, with the first AI factory scheduled to begin operations in 2027. 

The network relies on graphics processing units optimized for large-scale model training, post-training, inference, and agentic AI workloads.

The infrastructure uses the Nvidia DSX full-stack reference architecture. 

The deployment integrates Nvidia DSX MaxLPS software to regulate token processing throughput per megawatt, alongside the DSX operating system for multi-tenant platform lifecycle management. 

Through the agreement, SK Telecom joins the Nvidia Cloud Partner program, giving the telecommunications provider access to Nvidia's developer ecosystem and hardware updates.

The infrastructure project follows SK Telecom's recent integration of open-source Nvidia Nemotron datasets to train its domestic sovereign AI foundation model, designated as A.X K1.

Beyond the individual corporate agreements, Nvidia and SK Group announced a joint research initiative to design next-generation AI factory architectures. 

The joint research teams will focus on optimization projects spanning full-stack computing infrastructure, advanced memory technologies, data center energy grids, and operational resilience.