When Networks Start Making Their Own Decisions
Telecom networks are becoming too complex to manage through human intervention alone. 5G, cloud infrastructure, edge computing, IoT and increasingly software-defined networks generate enormous volumes of telemetry, alarms and service events that operators must interpret in real time. The challenge is no longer simply collecting more network data. It is turning that data into decisions quickly enough to act on it.
The industry is now moving toward a different model: networks that can detect problems, understand their causes, decide how to respond and, in carefully governed situations, execute those responses themselves.
For years, telecom automation has focused on predefined rules: if a certain threshold is breached, trigger a particular workflow. The next stage is more dynamic.
Agentic AI systems can combine network telemetry, topology, historical events and operational policies to reason about what is happening and determine the appropriate response. Google Cloud, for example, describes the emerging model as a shift from AI that provides insights toward agents capable of sensing, reasoning and taking action. Its 2026 autonomous-network work includes agents being trialled with One NZ that can independently reroute traffic or reset network settings when problems are detected.
The ambition is to reach higher levels of network autonomy, where systems can move from detecting → diagnosing → recommending → acting, while humans retain control over policies and higher-risk decisions.
That matters because the cost of manual network operations grows as networks become more distributed. NVIDIA estimates that telecom operators spent nearly $295 billion in capital expenditure and more than $1 trillion in operating expenditure in 2024, including spending on traditionally manual network planning and maintenance processes.
Company Spotlights
Future Connections – Building the Agentic Network Layer
Future Connections is a Netherlands-based telecom AI company focused on network intelligence, automation and assurance. Its portfolio includes AI-powered RAN applications and NIx Agents, designed to move network operations toward greater autonomy.
In 2026, the company has been particularly active in this transition. It partnered with Google Cloud on autonomous network agents, including a Core Network agent deployed with New Zealand operator One NZ. The company is also developing multi-agent orchestration and optimisation capabilities for telecom operators.
NetAI – Teaching AI to Understand Network Relationships
NetAI is taking a graph-based approach to network intelligence. Instead of treating network information as disconnected data points, its Graph Neural Network technology models the relationships between network elements to help identify the likely source of incidents.
In collaboration with Google Cloud, NetAI has been involved in a 2026 MasOrange proof of concept exploring GraphML-based AIOps. The approach is designed to give AI systems a better understanding of network topology and improve confidence when moving from diagnosis toward autonomous action.
Enfec – Building an Agentic NOC
Enfec is developing an Agentic NOC, designed to move network operations from traditional monitoring toward Level-4 autonomous operations. Its work includes network observability, AI/ML-based health monitoring and prediction, and automated incident management.
In June 2026, Enfec’s Agentic NOC work was recognised in the DTW Ignite Innovation Awards under the Autonomy Accelerator category. The company is working with telecom infrastructure at Tier-1 scale, making its approach an interesting example of how autonomous operations are moving from concept toward production environments.
What Comes Next?
The question for telecom operators is no longer whether AI can analyse network data. It is how much decision-making they are willing to delegate to it.
Get innovation insights from The FutureList weekly. Subscribe to our newsletter here