By Light Reading
Publication Date: 2026-03-11 15:34:00
T-Mobile is one of Nvidia’s biggest telco fans. Along with Japan’s Softbank, the US carrier has embraced the chip giant’s AI RAN idea that future radio sites could be transformed into edge nodes for AI inference with its graphics processing units (GPUs). It shows affinity by inviting Nvidia CEO Jensen Huang to speak at its latest strategy updates and sending its execs to appear at recent Nvidia events.
But T-Mobile’s AI RAN plan is not all about Nvidia. Like many carriers, it already uses AI in the RAN for performance and efficiency gains, which are uses typically described as AI for RAN. The carrier expects to inject more such AI in the network as it works toward a future RAN architecture.
Speaking to Light Reading at Mobile World Congress last week, Ankur Kapoor, T-Mobile’s chief network officer, shared how the operator is using AI in its 5G standalone (SA) network and his view that getting more AI in the RAN is a step on the way to AI RAN.
For example, in a disaster recovery scenario, cell sites can automatically adapt to provide mobile service where it is needed, such as providing coverage to first responders. This is T-Mobile’s self-organizing network capability that uses real-time AI and predictive data modeling to change where signals are beamed.
“That’s an AI-in-RAN implementation. The AI algorithms are already built into the RAN that we customized. Cell sites change the antenna tilt … [and other…