The rise of the autonomous network: How GraphML is redefining telecom operations

Modern telecommunications networks are becoming increasingly complex, requiring advanced management strategies. Communication Service Providers (CSPs) are now building autonomous networks to handle this complexity using AI and automation. Google Cloud's framework utilizes real-time digital twins and advanced graph machine learning (GraphML). GraphML, using graph neural networks (GNNs), analyzes and anticipates potential network issues. This approach is similar to how Google manages its own large-scale networks. Google partnered with NetAI to bring this technology to the telecom industry. MasOrange showcased a Proof of Concept (PoC) at MWC 2026 to explore fully managed AIOps. The digital twin models network relationships and topology, which is essential for understanding network behavior. Unlike traditional machine learning, GNNs understand network topology and relationships, leading to more accurate predictions. The Google tf-GNN library and NetAI's expertise combined to create tailored models for telecom scenarios. This collaboration aims to provide CSPs with the tools for truly autonomous and self-healing networks.