T-Mobile US and Ericsson moved a pair of AI-controlled radio access network (AI-RAN) services designed to improve 5G network efficiency into commercial trials across the carrier’s 5G-Advanced network, extending what has been accelerated work around AI-RAN initiatives.

The latest advance has the pair moving AI-native scheduler and link adaptation into “large-scale commercial trials” on live 5G-Advanced network traffic. This software taps neural networks running on Ericsson “optimized” hardware that can predict changing radio conditions in real time, with the trial showing a 10 percent improvement in spectral efficiency and 15 percent increase in downlink speeds compared with “legacy rule-based methods.”

Those stats matched previous technology trial results conducted in more limited conditions, “demonstrating the solution’s adaptability to diverse environments as the two companies jointly broaden the footprint further.”

“Our work with Ericsson on AI-native scheduler with link adaptation demonstrates how real-time, AI-driven optimization can enhance spectral efficiency and throughput while delivering a more consistent experience for customers at scale,” Grant Castle, SVP of RAN engineering and emerging technologies at T-Mobile US, noted in a statement.

AI-RAN work running

The work builds on long-standing work between the two companies and the AI-RAN community, with much of the collaborative work tied to the AI-RAN Alliance. That group was formed in early 2024, with a stated goal to steer the use of AI into RANs for better performance, lower operating costs, greater efficiencies, and to support new business models.

T-Mobile US was one of the AI-RAN Alliance's founding members and one of the first operators to get behind coordinated AI-RAN efforts when it launched a test facility in late 2024. The Washington-based facility is focused on testing the use of AI to help control RAN functions, and includes partners like Nvidia, Ericsson, and Nokia, all of whom are part of the AI-RAN Alliance.

John Saw, president of technology at T-Mobile US, during an earnings call earlier this year, explained that these efforts have so far evolved to both vendor partners Ericsson and Nokia having successfully made voice calls over an Nvidia-powered AI platform, but that there are much broader opportunities to be had with “physical AI.”

“That's why we see the potential there,” Saw said. “If you talk about tokens, the basic compute unit for AI is tokens … and with generative AI, tokens are usually just information tokens. They can be processed in a few seconds. It basically gives you a good analytics of the world. But with physical AI, tokens are going to be moving, they're going to take action. We coined it ‘kinetic tokens,’ tokens that move. When tokens move, when things move with AI, I think it gives telecom operators a license to play in a big opportunity with physical AI. It could be worth – if you believe some experts – hundreds of trillions of dollars as a total [addressable market].”

Tokens are units of data that are typically a piece of a word, or in the case of non-text modalities, a portion of an image or an audio clip. Tokens have now become the de facto billing metric for AI services, with enterprises paying by the number of tokens that their AI consumes.

That opportunity is also reliant on 6G, which, while still years away from an actual standard, is already starting to power hyperbolic potential.

“We believe that 6G is the connective tissue for physical AI … and it is going to be the intelligent fabric that connects data centers, network Edge, as well as AI devices itself,” Saw added. “So 6G is not just the connecting pipe, but really the nervous system for physical AI.”

Ericsson was also a founding AI-RAN Alliance member and has continued to build on AI-related RAN work, including being part of the recently launched Open Centralized Unit Distributed Unit (OCUDU) Ecosystem Foundation that sprang from the Linux Foundation late last year.

That work has roots in the US Department of Defense (DoD) to accelerate work on streamlining AI efforts to benefit 6G technology development, and involves “premier partners” such as AMD, AT&T, DeepSig, Nokia, Nvidia, SoftBank, SRS, and Verizon, with T-Mobile US linked as a “general” member.