- T-Mobile plans to start AI-RAN field tests in 2026 after doing joint trials with Nvidia, Nokia and Ericsson
- GPUs in the RAN? Nokia’s new AI-RAN solution will embed Nvidia GPUs – a controversial but potentially game-changing step toward AI-native networks
- Laying the groundwork for 6G: T-Mobile CTO John Saw says AI, cloud and 6G will converge – and T-Mobile’s 5G standalone network is ready for it
It’s been just over a year since T-Mobile revealed its work with Nvidia using AI in the RAN, so given this week’s news, we got to wondering: What kind of progress have they made since then and how close are we to seeing this all come to fruition?
Fierce caught up with T-Mobile President of Technology and Chief Technology Officer John Saw for an update. Saw, as you may recall, spearheaded Sprint’s rollup of its 2.5 GHz spectrum, which now serves as the mid-band anchor for T-Mobile’s 5G network. He’s also founder of the AI-RAN Innovation Center, where they’re working with Ericsson, Nokia and Nvidia to supercharge the RAN.
Earlier this week, T-Mobile said it will conduct field evaluations and testing of advanced AI-RAN technologies starting in 2026. In reality, a lot of tests have already occurred in collaboration with Nvidia, Nokia and Ericsson since last year’s T-Mobile Capital Markets Day, where a lot of this AI-RAN work was announced.
The tests are evaluating things like costs, performance capabilities and basic stuff like how much power is required for various tasks, such as real-time machine learning and methods for improving spectrum usage. Saw said he doesn’t know how long the testing phase will take – it depends on how the technology matures.
AI is the future
One thing is easy to predict: AI is the future. “The future of the network is going to be AI native,” he said. “What we mean by that is basically, the AI native stack is integrated into every layer of the network – hardware, software and the architecture.”
T-Mobile already uses AI to improve its network. For example, it uses self-organizing network technology powered by AI to steer antenna coverage to areas where customers congregate when there’s a natural disaster. When Hurricane Milton hit Florida, T-Mobile lost a lot of cell sites but only 8% of customer connections went down because it was able to steer antenna beams to the right areas, he said.
With AI in the network, T-Mobile and its partners like Nvidia are talking about potentially creating new, yet-to-be defined revenue streams in the future. The expectation is AI will result in faster, better performance for customers, as well as new applications that take advantage of the cloud and AI.
Nvidia, Nokia and the the big deal
As part of the partnership announced this week, Nokia intends to embed Nvidia Aerial RAN Computer Pro (ARC-Pro) into its new AI-RAN solution, enabling Nokia’s mobile network customers to transition from today’s more traditional RAN networks to future AI-RAN networks.
Essentially, they’re talking about putting graphics processing units (GPUs) into base stations.
That’s a somewhat controversial move in telecom circles, as some operators question whether the higher GPU costs are really necessary given their power requirements and the perceived lack of performance boost.
Fierce asked Saw about these reservations. He harkened back to the days when people told him 2.5 GHz was too high on the spectrum range to work for wireless.
“If I had listened to them, we wouldn’t be where we are today,” he said. “Sometimes when you’re trying to change a game and disrupt the technology that’s everybody’s used to doing, you will have skeptics out there. You just have to give it a chance. I think it’s too early to write off GPUs as too expensive.”
It’s worth noting here that T-Mobile hasn’t yet committed to buying a specific product, although one can surely see the future intent.
“We have been an active participant in helping to develop a new architecture and a new platform,” he said. “Obviously, if this works for the right performance and at the right cost points, we will be happy to deploy it, especially if it's based on a [Nokia] AirScale platform that we have all come to know and love.”
T-Mobile: We move fast
As for the speed at which all this is happening, telecom service providers generally have a reputation for moving at a snail’s pace.
Fierce asked about this – which gave Saw took the perfect opportunity to address T-Mobile’s pace, which arguably moved faster in the past five years than the prior 20 years.
“There's a reason why we are the 5G leaders that we are today. It’s because we move fast, and we move faster than our competition,” he said.
OK, we’ll give him that. T-Mobile’s nationwide 5G standalone (SA) network launched five years ago and earlier this year, it made the transition to nationwide 5G Advanced. AT&T and Verizon just recently announced nationwide 5G SA.
But T-Mobile isn’t doing all this alone. “We are very good at what we do, but we're not very good at other things that other companies are good at. So we are delighted to partner with Nvidia and work closely with them. These guys are world champions in AI processing,” he said.
That, incidentally, involves learning Nvidia’s language and vice versa – and none of this is happening overnight. They’ve been at this for more than a year, working with Ericsson as well as Nokia.
“This is how you create synergy and this is how you create disruptive capabilities that traditional telecom will never be able to see,” he said. “We believe in partnering with the best out there – with world champions who can take us to the next level.”
Saw: 6G is where it’s at
Some folks question the timing of AI in the RAN and 6G. The reality is, the AI-RAN Alliance launched at Mobile World Congress in Barcelona in 2024. By then, most operators were well down the path of deploying 5G and retroactively putting AI into the RAN didn’t sound like the best idea. With 6G standards being defined now, it’s much easier to bake AI into hardware and software.
Saw said the ability to combine AI, cloud and 6G at the same time creates an opportunity to make 6G into a technology that’s unlike anything else that’s come before.
5G SA is foundational to 6G. “Our network is born ready for 6G and that’s why we’re pushing initiatives like the AI RAN. We see where the future is going,” he said. “It’s going to be the intersection of AI and mobile networks.”