TLDR: China Mobile has successfully elevated its network operation center to Level 4 Autonomous Network capabilities, driven by the strategic implementation of intelligent agents and sophisticated AI models. This significant achievement has resulted in substantial operational efficiencies, including a reduction in manual labor equivalent to 5,500 full-time roles and marked improvements in fault resolution and customer service, earning the company a TM Forum Excellence Award.
China Mobile, the world’s largest mobile network operator by subscribers, has announced a groundbreaking achievement in network management, reaching Level 4 Autonomous Network (AN) capabilities within its network operation center (NOC). This advancement signifies a pivotal shift towards highly automated and self-managing network operations, largely attributed to the deployment of intelligent agents and advanced Artificial Intelligence (AI) models.
The company’s journey to Level 4 AN, as assessed by TM Forum’s AN Levels assessment (ANL), involved the development and integration of automated and intelligent agents and co-pilots specifically designed for operations and maintenance (O&M). This initiative was bolstered by significant investment in accurate telco generative AI (GenAI) and other AI models, alongside the strategic utilization of TM Forum’s established frameworks.
The impact of this technological leap is substantial and quantifiable. China Mobile reports that by the end of 2024, the intelligent agents had replaced the equivalent of 5,500 full-time manual roles. Operationally, the company has seen over a 30% reduction in backend O&M manpower and more than 5% savings in frontline installation and maintenance manpower. Furthermore, the Mean Time To Repair (MTTR) for faults and customer complaints has decreased by an average of 30%.
Specific improvements highlight the depth of this automation: 40% of core network back-end operations are now performed by digital employees, leading to an 87% reduction in core network fault MTTR and a 65% reduction in end-to-end complaint MTTR. The complaint pre-emption rate has dramatically increased to 75%, up from a previous 5%. In IP backhaul fault management, agents now handle 30% of all back-end staff tasks and 5% of field staff tasks, resulting in a 25% reduction in fault MTTR and a 15% decrease in fault tickets, with real-time self-service replacing interactions that previously took 30 minutes.
This achievement is particularly significant given the immense scale of China Mobile’s infrastructure. The company manages the world’s largest 5G network, serving 578 million 5G customers and over 1 billion total mobile subscribers. It also operates the largest 400 Gigabits per second optical transport network (OTN) and SRv6 backbone globally, in addition to the biggest telco-based, single-cluster intelligent computing center. By 2023, China Mobile had already reached AN level 3.2 across high-value network scenarios, setting the stage for its ambitious Level 4 target.
Level 4 autonomy represents a critical transition from human-defined automation processes to genuine autonomous decision-making. This requires end-to-end automation in high-value scenarios, particularly in fault management and customer complaint handling. China Mobile aims to extend Level 4 autonomy to almost all services by 2025, encompassing various domains including mobile RAN and core, IP, optical, and telco cloud networks.
The company’s broader ‘AI+’ strategy includes the deployment of 70,000 digital employees, which have cumulatively reduced workload by 7 million person-days. Supporting this is a robust AI infrastructure, featuring two mega 10,000-GPU intelligent computing centers in Hohhot and Harbin, and a ‘Computing Network AI Brain’ deployed across multiple national nodes, delivering a total intelligent computility of 61.3 EFLOPS (FP16).
Also Read:
- O2 Telefónica Forges Ahead in Agentic AI with Nvidia and Tech Mahindra for Network Transformation
- Aurora Mobile’s GPTBots.ai Revolutionizes Manufacturing with High-ROI AI Solutions
China Mobile has also leveraged Huawei’s Xinghe Intelligent Solution, which employs a three-layer architecture comprising intelligent network elements, digital twins, and a central cyber brain. This solution integrates a telecom AI model supporting 10 billion parameters and cloud-map algorithm simulations to establish an L4 ‘autonomous driving network’ capable of independently identifying risks, resolving faults, and verifying changes. Furthermore, the company has piloted Huawei’s universal intelligent service processing units (UISPs) on the radio access network (RAN) side, deploying over 80,000 intelligentRAN base stations. These units transform RANs into sensing and decision-making nodes, optimizing user experiences, simplifying O&M, and improving energy efficiency.
