Nokia Projects AI to Facilitate Nearly Half of Traffic by 2035

A new report from Nokia projects that artificial intelligence will account for 44% of global network traffic over the next decade. The Finnish manufacturer’s Global Network Traffic Report forecasts that total network volume will nearly quadruple from 2026 levels, reaching 3,789 exabytes per month by 2035.

Physical AI applications, such as autonomous vehicles and robotics, are expected to generate over 90% of their traffic via the uplink by 2035. Traffic from these physical systems is forecast to increase 105-fold, while agentic AI traffic is projected to grow 160-fold. Uplink traffic may represent 24% of total access volume in the most AI-concentrated network cells.

The evolving workload is projected to influence where AI processing occurs. While centralized cloud infrastructure will manage the majority of traffic, edge and metro core networks are expected to process 36% of AI-native traffic by 2035. Nokia attributed this shift to demand for lower latency and sovereignty-compliant models, moving inferencing closer to end-users and devices. Furthermore, fixed wireless access (FWA) is expected to represent more than one-third of consumer cellular traffic by the same period as adoption and usage continue to expand.

The vendor cautioned that changing traffic dynamics could stress existing infrastructure, particularly in high-usage areas, necessitating networks capable of supporting diverse requirements from bulk transfers to real-time connectivity. “AI is redefining network requirements, not just volume,” stated Pallavi Mahajan, Nokia’s chief technology and AI officer. She emphasized that network readiness must be central to AI strategy, as connectivity demands grow increasingly complex.