September 19, 2024
By Bob O'Donnell
It hasn’t been the easiest of times for 5G. After all, the technology that was supposed to change our world hasn’t exactly hit that lofty target. Instead, we’ve had to be content with some decent improvements in download speeds on our mobile phones and a new means of getting broadband internet access. Nice to have—to be sure—but a lot less than the overhyped predictions we’d hoped for.
At the company’s recent Capital Markets Day, however, T-Mobile made a more than respectable effort at reigniting some excitement around 5G and what it can do to enhance both our personal and professional lives. In addition, the company made it clear that it intends to stay in front of the latest technological enhancements in order to be seen as the technology leader among major service providers. Not surprisingly, much of its efforts are focused around or based upon AI.
What was surprising, however, was the kind of endeavors the company is embarking upon and with whom it’s partnering on these AI efforts. Both OpenAI CEO Sam Altman and Nvidia CEO Jensen Huang (separately) joined T-Mobile CEO Mike Sievert onstage to discuss these intriguing new initiatives.
The partnership with OpenAI is for a new customer service offering T-Mobile is developing called Intent:CX that is scheduled for next year. Labelled by T-Mo as an “intent-driven AI decisioning platform,” Intent:CX uses OpenAI’s platform and GPT foundation models to analyze T-Mobile’s enormous cache of customer usage data to mine it for insights. While that initially sounds a lot like one of the many GenAI-powered customer service offerings that we’ve seen other organizations introduce, T-Mobile claims that Intent:CX is expected to use some of OpenAI’s latest reasoning-based models (such as the new o1) to move beyond typical support capabilities and enable autonomous actions. The idea is to not only provide better and more personalized support through a combination of learnings from existing customer interactions, along with knowledge of T-Mobile offerings and more, but to also leverage real-time network interaction data and even make suggestions or perform automated actions on the customer’s behalf based on that data. Ultimately, T-Mobile believes it can reduce support calls by 75% while simultaneously improving the quality of the support—quite the goal.
From a business perspective, T-Mobile says that means they can reduce the number of support personnel needed, dramatically reduce the need for 3rd party call centers, reduce the number of retail stores required, and still improve customer support, increase customer satisfaction, and reduce churn. Needless to say, that’s exactly the kind of ROI for GenAI-powered applications that a lot of companies have been talking about, so it would be a great example of the revenue-saving potential of AI—if they can pull it off. However, given how notoriously bad much of the customer support typically is for service providers across the board, it’s not going to be an easy task.
There are also unanswered questions about how willing and accepting customers are going to be of these AI-powered customer support solutions—even with support personnel still involved. If the services are delivered seamlessly without most customers really noticing, then the impact could be quite positive. In fact, T-Mobile went so far as to say that its work could even serve as a blueprint for other industries with large customer support efforts that are hoping to leverage AI.
The partnership with Nvidia—which also involves network infrastructure vendors Ericsson and Nokia—initially entails creating a new AI RAN Innovation Center in T-Mobile’s headquarters in Bellevue, WA. The facility will be centered around an Nvidia GPU-equipped supercomputer running Nvidia’s Aerial software and connected to Ericsson and Nokia network equipment. As its name implies, the purpose of the facility will be to explore the possibilities enabled by enhancing T-Mobile’s RAN with AI.
Much of these efforts will focus on what Nvidia’s Aerial software can do. The purpose of Aerial is twofold: first, it’s designed to improve the performance of the Radio Access Network (RAN) that provides the core signal transfers at the heart of all mobile networks. Second, it also offers the ability to run AI workloads—both RAN and network core-related as well as completely separate types of efforts—within those mobile networks. In conjunction with the network infrastructure hardware from Ericsson and Nokia, the basic idea is that the Aerial software and AI RAN infrastructure can enable new opportunities for telcos to generate revenues.
Nvidia’s Aerial technology has actually been around for several years now. In fact, I first wrote about it in Forbes in June of 2021 (see “Nvidia’s Aerial Brings GPUs to AI on 5G”). From those early beginnings, it has always been an intriguing and potentially very impactful technology for two very distinct reasons. First, any efforts at optimizing and improving RAN performance lets mobile carriers like T-Mobile improve the speed, capacity, and efficiency of their networks, letting them carry more signals at a lower cost per bit. By training AI models about how the radio signals are propagated and feeding it a continuous real-time stream of data from the network, Aerial can automatically optimize network traffic for the best possible throughput. That’s a big win for network carriers like T-Mobile.
Second, while mobile carriers have long talked about their desire to bring more computing capabilities into their network and avoid being little more than dumb pipes, their real-world success on this front has been limited. By integrating GPU-powered computing capabilities into their networks, however—and essentially becoming AI-as-a-service providers—that’s a much more compelling story for telco operators to tell. It also presents a more realistic potential for T-Mobile to generate new revenues from these “smart pipes.”
The problem, up to now, is that Nvidia hasn’t had any luck in getting any of the major US (or European) carriers to make the significant investment necessary to integrate Aerial technology into their networks. Thanks to this new announcement, however, it’s clear that’s about to change. While T-Mobile didn’t provide an official timeline for moving this technology from the lab and deploying it across their network, the path forward has now been laid out and the presumption is that we’ll see some of these benefits starting later next year.
As important as this in-network AI computing news may be on its own, it’s even more interesting to analyze it in light of last week’s big announcement from Ericsson along with many of the world's largest mobile network providers, including T-Mobile, Verizon, AT&T, and many more (see my latest Forbes article entitled “Ericsson and Telcos Bringing Intelligence to Mobile Networks” for more details).
In case you didn’t see it, the basic gist of the news was that after years of talking about making networks an intelligent resource for mobile developers to leverage in the same way that they do computing and storage elements, the biggest players in the telco industry actually did something about it. Leveraging the efforts of the GSMA’s Open Gateway initiative and the CAMARA open-source network APIs, Ericsson was able to pull together major mobile providers around the world to support a new joint venture (whose name is still TBD) to create, support and monetize a global suite of network APIs.
In conjunction with today’s T-Mobile announcements, this implies that a company like T-Mo could eventually start to offer both the APIs and the necessary computing infrastructure to run those APIs within its network, thereby creating some very interesting new revenue opportunities. There are still a lot of details to be figured out to make this all happen, but it’s clear from all these announcements that after a long period of what appeared to be very few interesting developments, T-Mobile and its partners are working to bring some excitement back into the world of 5G.
Here’s a link to the original column: https://www.linkedin.com/pulse/t-mobiles-ai-efforts-make-5g-interesting-again-bob-o-donnell-q0rmc Bob O’Donnell is the president and chief analyst of TECHnalysis Research, LLC a market research firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on LinkedIn at Bob O’Donnell or on Twitter @bobodtech. |
