The core topic of MWC 2025 is undoubtedly artificial intelligence (AI), as the entire communications industry explores AI development trends from chips to end devices. Following SoftBank's announcement about the establishment of the AI-RAN Alliance at MWC 2024, although not much concrete progress was showcased this year, the number of alliance members has grown by nearly sevenfold.
On February 26, 2024, SoftBank announced the formation of the AI-RAN alliance, focusing primarily on three key areas—"AI-and-RAN," "AI-on-RAN," and "AI-for-RAN." Since its inception, the number of members has continued to expand, growing from an initial 11 companies to a global ecosystem encompassing 75 members across 17 countries within just one year. The number of service providers has also increased from 2 to 7.
Participating companies include leaders from the technology and telecommunications industries, such as equipment suppliers Ericsson, Nokia, and Samsung Electronics. Industry insiders believe that AI-RAN can provide strong momentum for 5G and Open RAN, laying the groundwork for AI-native 6G networks.
AI-RAN Alliance: progress and industry discussions
During the forum at MWC 2025, the chair of the AI-RAN Alliance and executives from founding members Nvidia and Samsung discussed the future development of "AI-Driven RAN Automation."
Alex Jinsung Choi, chair of the AI-RAN Alliance and principal fellow at SoftBank's Research Institute of Advanced Technology, pointed out that the concept of network automation has been a significant goal for the telecommunications industry for over a decade.
From software-defined networking (SDN) to network function virtualization (NFV), the industry continues to seek ways to reduce the total cost of ownership (TCO), which includes capital expenditures (CapEx) and operational expenditures (OpEx). However, challenges still hinder the advancement of network automation.
Despite years of active promotion of automation by telecom operators, issues such as infrastructure isolation and vendor lock-in persist. Traditional telecom networks consist of multiple independent domains, including transport networks, radio access networks, fixed access networks, and core networks.
Choi stated that while there has been some progress in automating specific domains, advancements in cross-domain and multi-layer orchestration remain limited. Additionally, vendor lock-in complicates collaborative operations across different network domains.
However, with advancements in AI and machine learning technologies, new breakthroughs are emerging within the industry.
AI as a catalyst for change
Choi believes that traditional AI has already been applied to network management. For instance, self-organizing networks (SON) are regarded as the first generation of AI applications in the telecommunications field.
Today, the emergence of generative AI (GenAI) and large language models (LLM) offers new solutions for multi-domain, cross-network, and multi-layer automation, enabling telecom operators to overcome existing limitations, enhance network performance, and reduce operational costs.
Industry experts generally agree that the application of AI in telecommunications infrastructure can be divided into two main directions. One focuses on network management and orchestration to improve overall operational efficiency, while the other applies to wireless algorithms for more precise spectrum and power optimization.
The combination of both will further drive RAN towards greater intelligence and automation, ultimately reducing TCO and enhancing operator competitiveness.
While the concept of self-organizing networks has existed for many years, its development remains relatively fragmented. Currently, the first step in network automation involves integrating agent technology with software virtualization, while further integration of AI—such as reinforcement learning, multimodal AI, and GenAI—offers greater potential for future development.
Article translated by Charlene Chen and edited by Jack Wu
