United StatesTechnology and telecom2026 – ongoing

T-Mobile US

AI-Managed Beamforming in live radio-network trials

T-Mobile reported approximately 16% higher downlink throughput during peak periods in New York beamforming trials

WorkingCompany's own claimOfficial ↗

Problem

Fixed allocation rules limited how the radio network assigned sounding reference signal resources as demand changed. Those resources helped antennas direct signals to devices.

Approach

T-Mobile worked with Ericsson on live commercial trials of AI-Managed Beamforming. A supervised-learning model trained on T-Mobile network data predicted demand and reassigned reference-signal resources. The model ran on Ericsson radio-network hardware.

Results

approximately 16%improvement in downlink user throughput during peak periods at New York trial sites, according to T-MobileCompany's own claim[1]

approximately 12%improvement in spectral efficiency during those periods, according to T-MobileCompany's own claim[1]

Takeaway

The reported gains covered peak periods at trial sites in New York City. They did not establish nationwide performance improvements.

Sources

5G Advanced in Action, Better Experiences: T-Mobile Puts AI to Work for Customers
T-Mobile20 Oct 2026Company publication

Source summaries

[1]T-Mobile20 Oct 2026

T-Mobile described supervised learning in live beamforming trials with Ericsson. It reported approximately 16% higher downlink throughput and 12% better spectral efficiency during peak periods in New York.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026