AI traffic reshapes telecom network planning and QoE solutions

Telecom operators are confronting a shift in network demand driven by artificial‑intelligence workloads. Executives from Telstra Digital Infrastructure described how sustained AI traffic of hundreds of gigabits, originating from an Australian cloud provider and destined for a U.S. network, reversed traditional traffic flows and forced the company to renegotiate interconnection capacity. They warned that agentic AI traffic requires guaranteed latency, unlike best‑effort training workloads, and that equipment lead times and subsea‑cable build schedules are tightening.

Broadcom’s VeloCloud division highlighted similar challenges for edge networks, emphasizing the need to identify AI traffic and optimise performance to meet quality‑of‑experience (QoE) targets. Product leaders outlined how their SD‑WAN platform can prioritize AI workloads, ensuring timely completion of inference tasks at the edge. Both sources see a move from selling raw infrastructure toward delivering guaranteed outcomes as AI workloads become pervasive across data‑centre corridors and eventually the edge.