NVIDIA announced on October 21, 2026, at GTC Berlin, that T-Mobile completed the first AI-RAN call on a commercial network, in Bellevue, Washington, together with Nokia. One classification matters: “world's first” is the companies' own description of the feat, confirmed by NVIDIA's official announcement and Nokia's blog — not an independent third-party verification.
AI-RAN stands for a radio access network (RAN) with AI-accelerated computing. In the Bellevue call, the network used commercial NVIDIA Aerial RAN Computer (ARC) servers, Nokia 5G anyRAN software accelerated by CUDA, Massive MIMO radios and an existing 5G Standalone core, with a commercial smartphone at the other end. In other words, the test left the lab and ran on a commercial cell in T-Mobile's production network.
Why it matters
The core idea of AI-RAN is to use the same accelerated platform for two jobs at once: running the radio network and executing AI workloads close to the user — such as voice, robotics and agentic applications — instead of keeping separate infrastructures. For operators, the argument is to use sites, power, fiber and spectrum they already own to offer distributed computing, not just connectivity.
According to NVIDIA, nearly 20 operators worldwide have already announced AI-RAN collaborations using Nokia's anyRAN and NVIDIA's ARC. In Europe, Elisa ran the continent's first field trial running voice, mobile data and video analytics at the same time on a single platform; BT plans live trials in Birmingham, England, in 2027; Deutsche Telekom validated the platform in labs in Bonn and Oulu and is also targeting field trials in 2027; and Orange is testing Nokia's 5G software on ARC.
New hardware and autonomous networks
Alongside the announcement, NVIDIA introduced two new ARC configurations, MGX 2U and MGX 4U, aimed at high-capacity centralized sites, with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs and x86 and NVIDIA Vera CPU options. Dell, Quanta Cloud Technology and Supermicro will offer ARC-ready servers. On another front, NVIDIA's open Nemotron models are entering network operations software: Nokia is integrating them into its Autonomous Networks Suite, and Cisco uses them in Crosswork AI, with TCS, BT and NTT DATA also adopting the approach for network agents.
The scale of the announcement should be kept in proportion: one successful call on a commercial network is a field-trial milestone, not a deployment at scale. Likewise, Nokia's goal of doubling spectral efficiency with AI-native RAN is an ambition stated by the company, not a result already demonstrated by this call. Commercial AI-RAN services, under the plans disclosed, still go through trials during 2027.
Sources: comunicado oficial da NVIDIA (via GlobeNewswire); blog da Nokia; Light Reading.
Image credit: official photo “NVIDIA-Powered Telecom Networks”, source NVIDIA Corporation, via GlobeNewswire.
Transparency: this article was produced with the assistance of artificial intelligence, based on public posts on X and online sources, including NVIDIA's official announcement and Nokia's blog. We recommend checking the original sources to confirm the details.
Muse
