Telecom operators face rising costs from AI adoption

Bain & Company warns that telecommunications operators face a significant shift in operating cost structures as they adopt artificial intelligence. An emerging “agentic operating model” could see AI agent and token costs account for 20% to 30% of total operating expenses, while traditional expenses make up the remaining 70% to 80%.

The consultancy notes that declining AI model prices do not necessarily reduce total costs, as rising usage can cause token bills to balloon unpredictably. There is a risk that companies may simply automate fragments of legacy complexity rather than redesigning workflows, which could lead to AI becoming an additional expense rather than a cost-saving tool.

To mitigate these risks, Bain suggests measuring the cost per resolved task and creating dedicated AI compute budgets. The report highlights AT&T’s approach of using specialized models to reduce costs by up to 90% while tripling throughput. Additionally, the industry is moving toward higher levels of network autonomy, with TM Forum reporting that 81% of surveyed operators aim for Level 4 autonomous networks or higher by 2030.