CSPs Adopt AI for Autonomous Transport Networks

Per a Cisco blog post that summarizes an Omdia study commissioned by Cisco and Ciena, researchers analyzed three-year roadmaps from 80 global communications service providers. The post reports 79% of CSPs now see AI services producing much higher traffic volumes, 61% are in early automation stages, and 47% already run AI in production networks. The study also reports 56% of interviewed operators expect their networks to operate autonomously or semi-autonomously within three years, and 64% plan to deploy digital twins next year to test AI-driven features before production deployment.

The study highlights three technical priorities: greater visibility, AI-driven forecasting and optimization, and closed-loop autonomy using agentic AI. Industry-pattern observations: service providers scaling AI workloads typically increase investment in telemetry, model-in-the-loop simulation (digital twins), and orchestration layers that can integrate model outputs into control planes. Those components are necessary to support rapid feedback and prevent model-driven instability in live networks.

For networking practitioners and ML engineers embedding AI into operational systems, the study documents an acceleration from manual CLI processes to automated, model-guided operations. Industry observations note that moving from advisory AI to closed-loop control raises challenges in validation, testing, change management, and safety engineering, especially when model decisions affect live routing and capacity.

Observers should track adoption metrics cited in the study: digital twin rollouts, the share of live production AI use cases, and the prevalence of closed-loop agent deployments. Also watch for technical disclosures about how operators validate model actions, rollback mechanisms, and standards for telemetry and model explainability in transport networks.