SoftBank advances AI-RAN with groundbreaking performance improvements, cost-saving solutions, and sustainability-driven partnerships, redefining next-gen mobile networks at MWC 2025.
SoftBank Corp. (“SoftBank”) led a series of major announcements this week that will drive the widespread global adoption of AI-RAN infrastructure.
Alongside significant enhancements to SoftBank’s proprietary converged AI-RAN solution, “AITRAS,” SoftBank and its partners announced a range of technological developments at MWC Barcelona 2025.
SoftBank has pioneered AI-RAN since the concept was first announced at its Annual General Meeting of Shareholders in June 2023 – followed in 2024 by the formation of the AI-RAN Alliance at MWC Barcelona 2024 and the subsequent launch of AITRAS in November 2024.
Significant momentum is now building behind AI-RAN as a way to transform the telecommunications industry and help it to address long-standing issues of connectivity and revenue generation.
Driving RAN performance and significant capex savings
One of the biggest achievements announced was SoftBank’s successful demonstration of RAN performance improvement using AI technology. In an initiative led by SoftBank and jointly conducted with NVIDIA and Fujitsu to enhance uplink (“UL”) channel estimation accuracy for signals received at base stations, the companies were able to improve UL user throughput by approximately 20% on smartphones. In another initiative, by predicting the timing of Sounding Reference Signals (“SRS”) transmitted from devices, SoftBank achieved an approximate 13% improvement in downlink (“DL”) throughput in high-mobility scenarios. In a third initiative, through AI optimization of MAC (MAC-layer radio resource allocation) scheduling in MU-MIMO (Multi-User Multiple Input, Multiple Output), SoftBank was able to improve average user throughput by approximately 8%.
Improvements of this magnitude would normally require significant levels of infrastructure investment. Deploying this on a large scale through the AI-RAN solution could help operators globally save billions of dollars on mobile infrastructure capital expenditure, while bringing new AI-related revenue sources to the industry and improving performance for end users.
Dynamically optimizing power consumption and renewable energy usage
A joint initiative between SoftBank and Red Hat has now enabled operators to intelligently monitor and optimize energy usage across their data center portfolios. The partnership enables monitoring of power usage for AI applications running on AITRAS by implementing Red Hat’s Kepler into the virtualization infrastructure of AITRAS. With this implementation, the AITRAS Orchestrator can dynamically allocate optimal resources across the network – for example, by moving AI workloads to where there are plentiful sources of renewable energy. This means network operators can save energy and accelerate the deployment of renewables, contributing to the decarbonization of the industry.
Groundbreaking collaboration between SoftBank and leading vendors
SoftBank and Nokia have developed a solution that enables the coexistence of AI applications and vRAN on a single server with GPU. Additionally, the two companies integrated Nokia’s EMS (Element Management System), MantaRay NM, with the AITRAS Orchestrator to allow for the optimal allocation of resources based on traffic behavior predictions derived from monitoring data, such as the number of connected users. This partnership with Nokia realizes the AI-RAN concept, enabling operators to use unused computing resources dynamically for AI inference. It enhances efficiency, lowers frequency investment, and improves AI services and mobile experiences.
This was followed at MWC Barcelona 2025 with the signing of a new MoU between SoftBank and Ericsson for joint research and development to advance AI-RAN. Extending a partnership signed in 2024, this new agreement includes jointly research on expanding the operation method of Ericsson Cloud RAN software running on the NVIDIA GH200 Grace Hopper Superchip to utilize GPUs, and conducting verification.
Source: https://techafricanews.com/
