As global digital and intelligent transformation enters a critical phase, the communications industry stands at a key inflection point, shifting from the connectivity era to the AI-native era. With the large-scale commercialisation of 5G-Advanced, accelerated R&D of 6G, and the maturity of new technologies such as AI large models and digital twins, autonomous networks (AN) are rapidly evolving from automation-assisted operations and maintenance (O&M) towards autonomous decision-making. This is more than a tool upgrade, empowering global operators to shift from cost reduction to revenue generation, writes Peng Zheng, the general manager of Service and Data Intelligence at ZTE.
According to TM Forum, more than 60% of the world’s leading operators have deployed TM Forum Autonomous Networks Project Level 3 (L3) autonomous capabilities. Pioneers including China Mobile, Deutsche Telekom, Vodafone and China Telecom have achieved L4 closed-loop in multiple high-value scenarios, signifying that autonomous networks have moved from concepts to large-scale deployment. By the end of 2026, both the proportion of operators meeting L4 deployment criteria and the global penetration rate of L4 scenarios will climb further. However, the excessively long static payback period (SPP) for L4 has become a major bottleneck hindering large-scale rollout. Reducing SPP has thus become one of the key factors in overcoming it.
Industry practice points to three breakthrough paths: first, standardising specifications to reduce development and integration costs and shorten deployment time; second, splitting complex workflows through refined process management; and third, tackling core technologies to enhance model accuracy. It is important to note that the key to advancing these paths lies in the fusion of three technical paradigms. As the cognitive brain, large models overcome the limitations of traditional rule engines by enabling natural language understanding, expert experience accumulation and long-horizon reasoning. As the execution unit, agents are goal-oriented with closed-loop perception, decision, execution and learning, completing complex tasks such as cross-domain fault self-healing. As the decision sandbox, digital twins provide a safe environment for validating AI decisions through simulation.
Based on the aforementioned technical paradigms, TM Forum standards and the practices of global leading operators, this article identifies three core development trends for autonomous networks over the next 3-5 years.
Peng Zheng