Nokia and AWS Teach 5G Networks to Run Themselves with AI Agents

Telecom networks are preparing for a radical shift: instead of an army of engineers configuring equipment by hand, AI agents capable of making decisions in real time will take over infrastructure management. This week Nokia and Amazon Web Services unveiled a pilot system in which artificial intelligence autonomously manages so-called network slicing in 5G networks, and it may prove far more significant than it first appears.

To grasp the scale of this, it helps to understand what network slicing is. Picture a highway carrying an ambulance, a freight truck and a cyclist at the same time. Each needs a lane with different characteristics: the ambulance needs priority and minimal delay, the truck needs steady throughput, and the cyclist is fine with a narrow path. Network slicing works on the same principle: a single physical 5G network is divided into virtual segments, each optimized for a specific task. One slice serves autonomous vehicles with ultra-low latency, another serves high-bandwidth video streaming, and a third serves Internet of Things sensors with minimal power consumption.

The problem is that managing these slices has so far remained largely a manual or semi-automated process. Engineers set rules, monitored load and reallocated resources, and inevitably lagged behind rapidly changing conditions. In a world where millions of devices generate unpredictable traffic patterns, this approach becomes a bottleneck. That is the problem Nokia and AWS are trying to solve.

Their joint system uses AI agents: not just machine learning algorithms, but autonomous software entities that can observe the state of the network, analyze changes and make operational decisions on their own. If video streaming load spikes in a particular district of a city, the agent shifts resources to the corresponding slice. If a business-critical industrial application starts to experience delays, the system reacts instantly, without waiting for an on-duty engineer to notice the problem on a dashboard. All of this happens in real time, at a scale human operators cannot match.

The choice of partners is no accident. Nokia is one of the world's three largest telecom equipment vendors, with solutions running in the networks of hundreds of operators. AWS is the dominant cloud provider, which in recent years has been expanding actively in the telecom sector by offering operators the chance to move some network functions to the cloud. Their alliance sets a precedent: for the first time, AI agents are gaining control of critical communications infrastructure not under laboratory conditions, but in a format close to industrial operation.

For the telecom industry this is a turning point for several reasons. First, economics. Carriers have complained for decades that they have become "dumb pipes": low-margin infrastructure companies while internet services take the profits. Intelligent slicing lets them sell not just connectivity but guaranteed service quality for specific applications, which opens up entirely new monetization models. Second, scalability. With the arrival of 6G and explosive growth in connected devices, manual network management will become physically impossible, and autonomous systems are shifting from an option to a necessity.

However, handing operational decisions in critical infrastructure to artificial intelligence raises serious questions. What happens if an AI agent makes a wrong decision and deprives emergency services of priority? Who is responsible if an autonomous system reallocates resources in a way that harms a critical service? Telecom regulators do not yet have a clear regulatory framework for such scenarios, and the Nokia and AWS pilot will inevitably have to confront the need to develop new standards for safety and accountability.

The broader context is also worth noting. The project fits into a global trend of deploying AI agents to manage complex systems, from power grids to logistics chains. Telecommunications has become one of the first industries where autonomous agents can demonstrate their value at industrial scale, simply because networks generate huge volumes of structured data and operate under fairly formalized rules.

The Nokia and AWS pilot is not a revolution that happened overnight. It is rather a first serious step toward a future in which telecom networks become truly self-managing organisms. And if the experiment succeeds, it will set the standard for the whole industry, defining exactly how artificial intelligence will be integrated into the infrastructure on which the connectivity of the modern world depends.