- T-Mobile and Ericsson are testing AI-managed beamforming on live network traffic in New York City
- T-Mobile reported a 16% increase in downlink throughput, with the strongest gains during peak demand
- The trials are part of T-Mobile’s broader push to bring AI deeper into the RAN ahead of 6G
Not one to miss an opportunity to plug its network prowess (and who isn’t?), T-Mobile is sharing results from AI-managed beamforming trials in New York City.
With Ericsson, T-Mobile deployed AI-Managed Beamforming in what T-Mobile Chief Network Officer Ankur Kapoor describes as large-scale commercial trials using live network traffic.
“This is another major step in bringing practical AI deeper into the RAN, building on our earlier AI-RAN work with Ericsson, including AI-Native Link Adaptation, and continuing the network vision we laid out at Capital Markets Day in 2024: a T-Mobile network that is not just bigger and faster, but more intelligent, adaptive and ready for the 6G era,” Kapoor said in a blog post.
Asked how many cell sites are involved, T-Mobile didn't say exactly, only that the initial trials are focused on "a targeted cluster of cell sites" in New York City, where they're evaluating the technology ahead of a larger deployment.
The capability is enabled through a network software upgrade and activated site by site.
"We plan to expand testing to additional markets later this quarter, building on the promising results we've seen in New York City," Kapoor told Fierce via email. "We're targeting a broader rollout across our markets in the first quarter of 2027."
Sizing up the results
Meanwhile, they're showing what’s possible with a 5G Advanced network and the right tools. Importantly, Kapoor said the strongest gains came during peak hours when customer demand is highest and the network must adapt quickly.
Here are some of the improvements they’re sharing:
- 50% increase in advanced beamforming using full-band Sounding Reference Signals (SRS)
- 16% improvement in downlink user throughput
- 12% improvement in spectral efficiency
- 10% increase in downlink data volume
According to Kapoor, these numbers are important because they point to a network that’s becoming more intelligent when customer expectations are highest.
“What matters most is what customers experience,” he wrote. “When demand is high, the network can respond more intelligently, putting capacity where it is needed and getting more from the spectrum we already have. The result is better performance when and where customers need it most.”
Ericsson’s role in all this
T-Mobile’s other major infrastructure vendor is Nokia, which famously received a $1 billion vote of confidence from Nvidia roughly one year ago. Given all that, it’s easy to get lost in all the fanfare about Nokia, which earlier this year made some impressive claims about the spectral efficiency gains it can achieve with AI-RAN.
Not to be outdone, T-Mobile's other Nordic vendor is getting some recognition of its own. According to Kapoor, Ericsson has been an important collaborator in bringing AI-Managed Beamforming into T-Mobile’s live network, combining its RAN and AI expertise with the scale and real-world demands of the network.
This latest development builds on the work T-Mobile did with Ericsson in AI-Native Link Adaptation, which reportedly led to an estimated 10% increase in spectrum efficiency and up to 15% boost in downlink throughput compared to legacy-based methods.
Now, enabled by Ericsson’s silicon and advanced network technologies, "this shared progress reflects our commitment to making networks more adaptive, more capable and ready for what comes next,” said Jaco du Plooy, SVP and head of Customer Unit T-Mobile at Ericsson Americas, in a statement.
To be sure, T-Mobile isn't playing favorites with any of its vendors. "We're working closely with both Ericsson and Nokia on a range of AI-RAN software innovations to make our network smarter, more efficient and more adaptive," Kapoor told Fierce. "Both partners play an important role in advancing our AI-RAN strategy and delivering better network experiences for customers."
T-Mobile and the bigger story
According to T-Mobile, it's all about learning what it takes to build a more intelligent, adaptive network that better serves customers.
Kapoor also reiterated something T-Mobile CTO John Saw has said recently: that 6G is more than just “another G” in the wireless industry’s march to the next generation of technology.
“6G will be about more than a new air interface or another speed milestone. Networks will need to sense, learn and respond as they support increasingly demanding applications, from distributed AI and immersive experiences to robotics, vision AI and other forms of physical AI,” Kapoor said.
“We are building those capabilities now … AI-Managed Beamforming is one step on that path,” he said.
Read more about AI RAN on Fierce Network:
T-Mobile fetes AI RAN innovation with Ericsson
Nvidia GTC: T-Mobile and Nvidia push physical AI to the network edge
The Five Nine: AI-RAN, 6G and the future of mobile networks