AT&T and Verizon move AI in the RAN toward scale tests

Speaking at the Telco AI Forum 2026, AT&T and Verizon executives described AI-based RAN deployments already delivering measurable gains. AT&T is trialing Ericsson's AI-native link adaptation with throughput and spectral efficiency improvements, while Verizon combines centralized rApps with local AI in the gNodeB. Both operators apply an "AI worthiness" test before scaling.

• AT&T trials Ericsson AI-native link adaptation with significant throughput gains
• Verizon pairs centralized rApps with real-time AI in the gNodeB
• Verizon requires AI gains to beat rule-based baselines at controlled cost
• Scaling demands open architectures, automation and model lifecycle management