ZTE Focuses on AI for RAN as Demand for RAN Computing Remains Limited

ZTE Corporation recommends that operators distinguish between using AI to improve radio access networks and using RAN computing resources for general AI applications, according to a sponsored article published by RCR Wireless News. The company favors a decoupled architecture because demand and business models for running third party AI workloads on RAN infrastructure remain limited.

ZTE reports that AI powered Massive MIMO can increase cell capacity by 15% to 20%, while AI based energy controls can reduce power use by 10% to 15%. AI supported operations can also reduce mean time to repair by 20%. The company estimates that differentiated service quality could increase average revenue per user by 5% to 20%.

A collaboration with China Mobile at a crystal marketplace reportedly benefited about 10,000 streamers within six months, by allocating network resources around service demand. Tests in Thailand and Indonesia found that users in guaranteed service scenarios were 20% more likely to receive speeds above 5 Mbps than other users, without affecting background traffic.

ZTE is also developing an Agent Team system in which specialized AI agents coordinate network optimization, fault management, energy controls and operations. RAN computing for third party AI inference remains on its roadmap while the company evaluates potential business models.