The impact of AI on wide area network traffic: we need to talk

Cisco has published a research report examining how AI—particularly agentic AI—is reshaping wide area network (WAN) traffic patterns. Based on live production network data and controlled experiments, the report finds that AI inference traffic behaves fundamentally differently from standard web traffic in terms of shape, symmetry, latency requirements, and criticality. Key projections include AI inference representing 25% of all network traffic by 2035, with the sharpest growth between 2029 and 2032, and a 63% additional traffic growth driven by AI's multiplying effect. The report argues that traditional network planning assumptions are no longer valid and calls for updated infrastructure strategies, including differentiated QoS, higher resilience, and continuous measurement frameworks for network leaders and service providers.