ZTE urges separation of AI for RAN and RAN for AI strategies
ZTE distinguishes between AI for RAN, which improves radio network efficiency, and RAN for AI, which uses network compute for general AI applications, with the latter lacking a mature business model. The company claims AI-powered Massive MIMO can boost cell capacity by 15-20% and AI-based energy saving can cut power use by 10-15%. ZTE advocates a decoupled, heterogeneous architecture to maximize return on investment.
ZTE's dual-layer intelligence architecture operates at both site and network levels, with the vendor reporting measurable gains across four operational areas. Field collaborations in Asia—including a crystal marketplace deployment with China Mobile and experience-guarantee trials in Thailand and Indonesia—demonstrate how the approach translates into real-world service improvements for livestreamers and event attendees.
The company's positioning reflects a broader industry pivot toward experience-centric monetization, moving beyond flat-rate data plans. By keeping AI-for-RAN optimization separate from general-purpose RAN compute, ZTE aims to address the paradox of surging data consumption against stagnant per-user revenue, while preparing infrastructure for future AI-driven bandwidth and latency requirements.
This distinction could influence how telecom operators allocate capital in the coming years. If ZTE's efficiency claims hold at scale, consumers may see improved network reliability and lower energy-related costs, while operators could gain new revenue streams through tiered service guarantees. However, the immature business case for RAN-based general AI suggests near-term societal benefits will likely center on network quality enhancements rather than novel AI-enabled services delivered through telecom infrastructure.