Samsung, Nvidia target vRAN spectral efficiency

At MWC26 Barcelona, Samsung and Nvidia outlined joint work on virtual RAN (vRAN) aimed at squeezing more capacity out of existing mobile spectrum, rather than relying on new airwaves. The companies tested Samsung’s vRAN software with Nvidia’s ARC Compact computing platform at a Samsung R&D site in February, confirming it runs in a live-style environment.

They plan to show how AI tools can raise spectral efficiency and support heavier network traffic using current spectrum holdings. The partnership also focuses on faster links between CPUs and GPUs using Samsung’s vRAN and a unified Nvidia processor for quicker data exchange. Nvidia’s VP of AI and Telecoms, Soma Velayutham, framed the work as answering operator demand for “AI-native, software-defined infrastructure,” while Samsung Networks strategy lead Keunchul Hwang said AI is becoming central to handling rising network loads.