SoftBank and Ericsson Test AI in RAN on Commercial 5G

SoftBank Corp. and Ericsson have conducted a pioneering validation of AI-native technology within the radio access network (RAN) on SoftBank's 5G commercial network, marking a significant step in the evolution of AI-enhanced mobile networks.

AI-native Scheduler Validation

SoftBank and Ericsson's collaboration tested the AI-native Scheduler for Link Adaptation, a feature of Ericsson's AI in RAN software, on SoftBank's 5G network. The validation demonstrated improvements in spectral efficiency and user throughput, with gains of up to 25% and 50% respectively, compared to conventional technology. This initiative is part of a broader strategy to enhance network performance and efficiency through AI integration. For more details, visit the official press release.

Commercial and Operational Implications

The successful validation of AI in RAN technology has significant commercial implications. By optimizing network performance in real-time, operators can better manage increasing traffic volumes and complex connectivity demands. This capability is crucial as the adoption of AI-driven applications and services continues to grow. The collaboration between SoftBank and Ericsson highlights the potential for AI to transform network operations, offering a competitive edge in the telecom sector. Insights into this development can be found on TelecomTV and SDxCentral.

Technical Collaboration and Roles

SoftBank and Ericsson's partnership involved distinct roles: SoftBank defined evaluation requirements based on traffic characteristics, while Ericsson trained and implemented the AI model. The collaboration underscores the importance of combining operator insights with technological expertise to achieve significant advancements in network performance. This partnership aims to further refine AI-native RAN software, aligning with SoftBank's operational environment. For additional context, see TelecomsTechNews and TelecomLead.

Limitations and Areas for Further Verification

While the validation results are promising, further verification is needed to assess the scalability and long-term impact of AI-native RAN technology. Future evaluations should consider diverse network environments and traffic conditions to ensure consistent performance improvements. Additionally, the integration of AI in RAN raises questions about data privacy and security, which require careful consideration and regulatory compliance. For more insights, refer to IBTimes Japan and TechBlog.

Future Prospects and Industry Impact

The successful implementation of AI in RAN technology positions SoftBank and Ericsson at the forefront of telecom innovation. As the industry moves towards 5G-Advanced and 6G, AI-native networks will play a pivotal role in supporting next-generation services. This development aligns with broader trends in AI infrastructure financing, such as NVIDIA's AI infrastructure financing and AMD's data center expansion. The collaboration also complements strategic partnerships like Oracle and AWS's AI database initiative and Ryanair's AI partnership with Google Cloud. For related SoftBank initiatives, see their February 2026 press release and January 2026 press release. Further industry insights are available on TelecomTV and NVIDIA Omniverse's advancements.

Operational evaluation of AI in RAN requires rigorous testing to ensure seamless integration with existing infrastructure. Decision-makers should scrutinize governance frameworks to address ethical and security concerns. Limitations such as data privacy and algorithmic bias must be acknowledged and mitigated. Procurement processes should prioritize transparency and vendor accountability. Validation of AI models is crucial, demanding robust methodologies to verify performance and reliability. An informed decision-maker should verify the alignment of AI initiatives with strategic objectives, ensuring that the deployment enhances network efficiency without compromising user experience or security. Continuous monitoring and iterative improvements are essential to adapt to evolving technological landscapes.