Verizon deploys AI agents to automate vRAN operations at scale

Verizon is moving beyond traditional automation by deploying AI agents to manage operations across what may be the world's largest virtualized radio access network, with early results showing promise for handling complex network tasks at unprecedented scale.

The carrier operates 22,900 vRAN sites—representing 40% of its total network as of early 2025—all running on its proprietary Verizon Cloud Platform (VCP). This on-premise infrastructure hosts both virtual and cloud-native network functions, including the carrier's 5G standalone core and vRAN, plus GPUs dedicated to network AI workloads.

Umashankar Velusamy, Verizon's senior director of technology development for network automation, outlined the company's approach at Network X Americas last month, according to Light Reading. He described three categories where AI agents are being deployed: planned changes like deployments and upgrades, unplanned responses to network degradation, and continuous optimization through RAN intelligent controllers.

Automation at unprecedented scale

Verizon has already automated vRAN deployments end-to-end once hardware is installed at the far edge, and can upgrade thousands of sites simultaneously. Omdia analyst Gabriel Brown noted that while many operators have automation pipelines for 5G core networks, few match Verizon's vRAN automation scale.

AI agents are now enhancing these capabilities. For planned changes, agents generate and validate configurations for new deployments and mass upgrades. In service assurance, they correlate data from multiple sources to accelerate problem resolution. The optimization category shows particularly strong potential, Velusamy said, with autonomous agents monitoring networks, identifying opportunities, converting them to actions, and verifying intended results.

Security and transparency requirements

Despite "good results" from current implementations, Velusamy emphasized the need for careful deployment with proper security guardrails. Converting domain knowledge into actionable context and making deterministic automations available as tools for agents requires significant care, he said.

The carrier is calling on suppliers to prioritize two areas: transparency and integration. "When we have agentic solutions being deployed, traceability and transparency are key," Velusamy said. Operators need to understand the reasoning behind agent decisions and the exact actions taken.

He also advocated for a common framework around security, trust, handoffs, negotiation, and context sharing—while maintaining that implementation should remain flexible. The AI ecosystem "is evolving way faster" than traditional telecom standards, he noted, requiring carriers to maintain agility in their independent work.

Why it matters

Verizon's deployment represents a test case for whether AI agents can manage telecom infrastructure at carrier scale. Success could accelerate the industry's shift from rule-based automation to adaptive systems that handle exceptions and optimize continuously—potentially reducing operational costs while improving network performance. The carrier's call for transparency and security standards also signals that agentic AI in critical infrastructure requires different guardrails than consumer applications.

Brown observed that Verizon's automation leadership could create a sustainable competitive advantage in network operations that ultimately improves customer experience.

These details were first reported by Michelle Donegan at Light Reading.