Digital twins are the key enabler of trust in agentic decision-making in CSP networks

There is a high level of interest in agentic AI in telecoms. However, putting the improvements in customer care chatbots to one side, the implementation of agentic AI in the operations of the network seems still a way off. Preliminary results in our upcoming agentic AI tracker show limited engagement beyond science projects and proofs of concept.

Appledore believe that a new generation of network-wide digital twins will form the foundation for trusting agentic AI across the organization. Today more ambitious network-wide and real-time digital twins are possible enabled by cloud scale, analytics/AI capabilities, and increasingly powerful real-time observability of the real network.

After a recent hiatus in interest, much of the renewed interest in digital twins in telco is being driven by the need to support AI and, in particular, agentic AI with robust data for effective decision-making. Digital twins provide a highly realistic model of the telco network to act as a starting point and sandbox to support decision-making. Agentic AI brings new relevance to digital twins. Appledore expect this interest in network digital twins to increase rapidly in the next two years, driven by the increasing need for opex savings in running networks and an increasing network skills shortage driven by retirement and competition from other sectors. Digital twins are the basis on which network operations “memory” can be retained and improved. Digital twins are the basis on which organizational memory can be scaled.

In this report, Appledore examines the wide range of possible agentic AI use case types that digital twins can support; use cases that we believe can stimulate growth in the Network Automation Software market over the next few years, particularly in Data, Inventory, and Observability Platforms. This reflects a shift in CSP investment focus (which has remained static for a decade) from network infrastructure to operational efficiency.