T-Mobile’s Ankur Kapoor: AI is moving from the cloud into the core network

I’ve been covering telecom networks since before there was an internet (seriously, just me and Peter Jennings, sharing a cube). And every so often, a carrier manages to change the trajectory of the industry. Over the past few years, T-Mobile has done exactly that. What started as a bold and expensive bet on a nationwide 5G standalone (SA) network — deep mid-band spectrum, cloud-native architecture and a very different view of how 5G should scale — has evolved into what is now widely regarded as the leading mobile network in the United States, including recognition in the latest JD Power U.S. Wireless Network Quality Performance Study.

One of the key architects behind that transformation is Ankur Kapoor, T-Mobile’s chief network officer. Kapoor has spent more than two decades with the company and has played a central role in building the network strategy that moved T-Mobile from an industry challenger to the operator many competitors are now trying to emulate.

Today, the focus is shifting again. With its standalone 5G infrastructure in place, T-Mobile is starting to embed artificial intelligence directly into the network itself — moving AI from the cloud toward the core and the edge. That shift could open the door to new consumer services, enterprise applications and a more intelligent, programmable network architecture.

I spoke with Kapoor about how AI is being integrated into the network, why edge intelligence will be critical for the next phase of telecom innovation, and how T-Mobile is approaching one of the biggest technical challenges facing operators: orchestrating increasingly complex networks across multiple vendors, domains and cloud platforms.

Building the foundation: 5G standalone 

Steve Saunders: Ankur, nice to see you. In a few words, what are the big technology trends that are really impacting T-Mobile and helping you move your business forward at the moment? 

Ankur Kapoor: That’s a great question. And I couldn’t be more excited to be talking about it. If I start with the very recent announcement we made at our Capital Markets event, that is one of the biggest inflection points for the industry. We have actually embedded AI into the network. It’s built into the core of the network now.

We talked about live translation, but live translation is just a use case. It’s not the end — it’s really the start of an era. The platform that we’ve established, with AI being in the heart of the network and the core network, is what makes it exciting.

But this was not just dumb luck. This is something we’ve been planning for years. The real start was almost six years ago when we began the path toward a 5G standalone network, a strategy T-Mobile outlined when it launched the nation’s first nationwide standalone 5G network. 

That really is the backbone of this entire platform. We had a vision that we needed three things: a 5G standalone network, deep mid-band spectrum and a cloud-native architecture. 

In the early days of 5G the focus was on millimeter wave deployments in stadiums and airports where lots of people were. But we had the ability to look around the corner. We believed 5G would be for everyone. 

If I had to define the T-Mobile network in a few words today, it’s scale, consistency and a platform.

From speeds to experience 

Saunders: It’s very significant that T-Mobile became America’s best network, not just its best 5G network. But let me go back to something you mentioned earlier — AI in the core of the network. What exactly is it doing there? 

Kapoor: If you think about the early days of 5G, or even 4G, the focus was all about speeds. The big shift that has happened in the industry is moving from speeds to actual experience. 

What we have built is AI that operates in the network core and is native to voice calling. It works on any device — even flip phones — not just smartphones or high-end devices. It also works globally. As long as one person is a T-Mobile customer and connected to LTE, 5G or even Wi-Fi, the service works even if you’re traveling across the world.

That’s what having AI in the core does. 

Early 5G conversations focused on speed and coverage. Now the focus is intelligence and adaptability, an approach the company is exploring further through initiatives such as its AI RAN innovation work.

The rise of edge computing and physical AI 

Saunders: One of the really exciting areas for me is what happens beyond consumer connectivity — particularly at the edge and in industrial environments. Is that something T-Mobile is focusing on as well? 

Kapoor: Absolutely. The shift from consumer use cases to enterprise use cases is very important. We’ve always said that 5G Advanced is the bridge to 6G. 

Looking ahead, one of the biggest developments is what we call physical AI. If you think about physical AI, it’s about autonomous networks, robotics and many other enterprise applications.

We’re very well positioned for that with our nationwide 5G Advanced network, the massive mid-band capacity we have and the cloud-native, AI-ready architecture we’ve developed.

We recently launched a product called Edge Control. What it does is bring processing closer to the customer. It brings computing directly to the facility where the work is happening. Enterprises don’t want to manage the control plane of a telecom network. What they want is for their data and their applications to run close to their operations.

That’s what Edge Control enables. It allows decisions to be made locally while we provide the technical depth and the connectivity — part of the company’s broader push into enterprise connectivity and private 5G services.

Solving the orchestration problem – exclusively  

Saunders: Let’s talk about orchestration. It seems to me that orchestration is becoming the key component in carrier networks. The problem is that every vendor has its own orchestrator. Nokia has one, Ericsson has one, hyperscalers have their own control planes. As someone who has to make all of this work together, how do you tackle that? 

Kapoor: You’re absolutely right — orchestration is one of the biggest challenges. Today every vendor is pitching some form of orchestrator, and they’re often organized by domain. You have radio orchestrators from companies like Nokia, Ericsson and Samsung. Then there are core orchestrators designed to connect the core network to the radio network. And then you have hyperscaler orchestration platforms. 

We have actually been working on this challenge for about six months. 

We now have a live deployed orchestrator that performs orchestration across the entire network — from the core network to the radio network — and it can even push over-the-air upgrades to devices. It’s not fully scaled yet, but it works, and it’s vendor-agnostic.

We’ve never believed in locking ourselves into one technology or architecture. We are architecture-agnostic. That philosophy has been central to T-Mobile’s success. That’s what this orchestrator enables.

Toward programmable networks 

Saunders: That’s breakthrough technology. Because orchestration can be internal — automating processes inside the network — but where it really matters is externally, where it connects to customers and services. In the future, customers should be able to choose exactly the network service they need. 

Kapoor: Exactly. My vision is that if you have a credit card and a specific requirement, that should be enough. You should be able to go to the T-Mobile website, create your own network slice, define your parameters and specify what you need — whether that’s low latency, higher download speeds or heavy uplink usage. If you’re broadcasting something or uploading lots of video content, you should be able to define that service yourself. 

That level of customer empowerment can only happen with unified orchestration. That vision is closely tied to capabilities emerging around 5G network slicing. We’ll still have domain-specific orchestrators — those are important for running the network — but what we really need is a master orchestrator that brings everything together. 

Recognition from customers 

Saunders: This has been fascinating. Thank you for sharing these insights, and congratulations again on the award for America’s Best Network.

Kapoor: Thank you. It feels really good — especially because the recognition from JD Power comes directly from customers. That means a lot.

Saunders: Enjoy it. The telecom industry can be full of naysayers until someone has real success. Now everybody wants to talk to you. You’re probably much more humble about it than I would be — but that’s why you’re in your job, and I’m sitting here in my kitchen. So thank you again. It’s great to see you.

Kapoor: You’re doing fantastic, Steve. Thank you.

For more of this kind of thing, click here.