T-Mobile has completed a 5G call using NVIDIA and Nokia’s AI-RAN technology on its live commercial network in Bellevue, Washington. It happened at an operating cell site with a commercial smartphone, rather than in a lab.
AI-RAN brings accelerated computing into mobile networks, with plans to run AI applications alongside radio processing. Network performance matters when you’re playing games through the cloud, especially over a mobile connection. We’ve already seen measurable improvements with Deutsche Telekom’s GeForce NOW network upgrades, and this is another 5G development we’ll be following.
T-Mobile Tests AI-RAN on Its Live 5G Network
The call used Nokia’s Massive MIMO radio equipment and 5G anyRAN software with NVIDIA’s Aerial RAN Computer platform. It connected through T-Mobile’s existing 5G Standalone core.
Nokia calls it the world’s first AI-RAN call at a commercial cell site. That’s the difference from earlier demonstrations. This time, the technology was connected to an operating mobile network, not tested only in a controlled environment.
The radio access network handles communication between your phone and the wider mobile network. With AI-RAN, operators can use GPU-accelerated computing for radio processing and potentially for AI applications at the same network locations.
Those shared AI workloads are part of the next phase of testing. Nokia plans to introduce more AI-powered radio functions and test AI applications alongside the network software.
AI-RAN Trials Expand Beyond T-Mobile
T-Mobile isn’t the only operator involved. In Finland, Elisa reports it has completed a live-network trial handling voice calls, mobile data and video analytics on the same accelerated computing platform.
Deutsche Telekom has carried out laboratory testing in Bonn and Oulu, with field trials planned for 2027. BT is preparing a live-network trial in Birmingham next year. Orange is also testing Nokia’s 5G software on NVIDIA’s Aerial RAN Computer platform.
Supermicro’s role is to provide servers that operators can install at different points in their networks. Its compact ARC-Pro servers combine NVIDIA Grace CPUs with RTX PRO 4500 Blackwell Server Edition GPUs for cell-site deployments. Larger ARC MGX configurations use RTX PRO 6000 Blackwell GPUs at network aggregation sites and central hubs.
Some NVIDIA ARC servers are available now. Supermicro expects the dual-Vera ARC MGX 2U later in 2026. ARC-Pro is scheduled for customer testing this year, ahead of general availability in the first half of 2027.
That gives operators more options for where to put computing capacity, including locations closer to people using the mobile network.
What This Could Mean for Cloud Gaming
Deutsche Telekom’s 5G+ Gaming service in Germany is a good example of how mobile network improvements can benefit cloud gaming. It works with GeForce NOW using L4S and network slicing to help manage gaming traffic.
In April 2026, Deutsche Telekom reported 20% lower latency and 73% fewer frames affected by packet loss for GeForce NOW connections receiving its gaming network treatment. The comparison covered October 7, 2025, through February 1, 2026, against connections without that treatment.
Those improvements came from L4S and network slicing, not AI-RAN. NVIDIA and Nokia are working on a different part of the network by bringing accelerated computing closer to where mobile connections happen.
That’s why we’d be interested in seeing an operator test cloud gaming on AI-RAN and publish actual latency and packet-loss measurements. We’d finally have something to compare against the mobile network improvements already being used for cloud gaming.
As always, remember to follow us on our social media platforms (e.g., Threads, X (Twitter), Bluesky, YouTube, and Facebook) to stay up-to-date with the latest news. This website contains affiliate links. We may receive a commission when you click on these links and make a purchase, at no extra cost to you. We are an independent site, and the opinions expressed here are our own.
