Samsung Electronics has completed a multi-cell test at its R&D centre, validating the use of its virtualised RAN (vRAN) software in combination with NVIDIA’s accelerated computing platform. The test represents a step towards commercial deployment, improved connectivity, and network performance.
Samsung plans to present this work at Mobile World Congress (MWC) 2026, with a focus on integrating AI enhancements into its network portfolio using NVIDIA’s platforms. One demonstration will feature AI-RAN technology, which uses AI algorithms to improve downlink performance, throughput, and spectral efficiency.
As AI technologies are increasingly incorporated into network operations, Samsung’s vRAN architecture remains a central component. Working with NVIDIA, the company aims to provide operators with enhanced flexibility and performance through CPU and GPU integration.
Both companies are exploring solutions to improve high-speed connections between CPU and GPU, including a unified processor that combines the components in a single chipset. This integration is intended to facilitate fast data exchanges and support efficient performance relative to total cost of ownership.
The collaboration recently included integration of Samsung’s vRAN software with NVIDIA’s ARC Compact system, which features the NVIDIA Grace CPU and L4 GPU. Samsung continues to investigate these synergies to develop network solutions through ongoing research and development.
With experience in large-scale vRAN deployments and computing partnerships, Samsung provides end-to-end 5G solutions, including chipsets and core networks, while exploring AI and automation applications in network operations.
