AT&T’s network plans: Touch every site, open up and optimize

‘We’re going to touch every cell site that we’ve got’, said COO Jeff McElfresh on AT&T’s network plans

AT&T laid out network plans earlier this week for a massive, three-year network modernization program that includes both plans to nearly double its fiber passings and step away from legacy copper infrastructure, as well as new details about the company’s push toward open Radio Access Network technology.

As its executives explained this week, the company’s playbook that has delivered solid growth remains the same: Focus on fiber and 5G, and delivering converged network services with owner’s economics at scale.

Jeff McElfresh, AT&T’s COO, provided additional details on what network modernization actually looks like for AT&T’s network plans during a press briefing as part of this week’s announcements.

First, the choice of Ericsson as the lead and foundation upon which AT&T’s open network will be built—an announcement which, at the time, seemed to fly in the face of what Open RAN had hoped to achieve in terms of disruption of the old telecom NEM order. Ericsson, McElfresh pointed out, provided the largest share of AT&T’s RAN.

“So it wasn’t so much that we said Ericsson is going to do this better than another provider. It was much more the case where their footprint in our network was significantly larger, and therefore the transformation had a little bit better headwinds [by] leaning in with them,” he explained. And, he added, AT&T put on the pressure for Ericsson to support the open direction that it wanted to move its network in. “We required them to open up their radio architecture—which, up until the deal that we drove, they were somewhat unwilling to let some disruptive technology partners into the mix,” he said.

Now, AT&T has announced two other radio vendors who will be in the mix, specifically for radios in urban areas: Fujitsu and Mavenir. There is room for other vendors, McElfresh said, but those two were “a little bit more mature” in what they had to offer today.

But as McElfresh said, the deal that is being called “the O-RAN deal” is actually about a much larger strategy, one that delivers an AT&T network which not only can integrate different types of open radios where needed, but is energy- and spectrum-efficient, digitalized and offers consistent capabilities that can be utilized by developers.

“We’re going to touch every cell site that we’ve got in AT&T’s footprint,” he said, adding that as AT&T does so, it will be creating digital twins of each site, using technologies such as drones and artificial intelligence, in order to insure that it has the right network assets at the right locations and right height. Touching each site will also enable AT&T to assess and streamline its power configurations and “clean up radio debt” in a way that it hasn’t really done in the past decade, in order to achieve new levels of network efficiency at scale, McElfresh said.

As it does that work at every site, he continued, AT&T will do so “with a head-nod toward open,” so that AT&T’s towers will either directly support open radios on the macro towers, or support open small cells in their serving areas as part of densification of the network, and everything will interconnect with the cloud RAN serving that particular geography.

While AT&T announced the shift toward open networks a year ago, McElfresh said that much of the year was spent testing and refining its processes for network swaps so that the work doesn’t disrupt network service significantly. Upgrades will ramp up quarter by quarter now, he said, and the work is expected to be complete close to the end of 2026 or early 2027.

The upgrades are happening not geography-by-geography, but in what AT&T calls a “popcorn approach” where towers and sectors across the country are being upgraded simultaneously. AT&T has already done “thousands” of sites to prove out and fine-tune its processes, McElfresh said, adding that it is “moving a lot more smoothly than I would have anticipated at this stage of the program.”

The vision, ultimately, is to have a network that can both easily integrate new, open equipment but also provide a platform for developers to innovate on. And the carrier also is thinking hard about how it will handle anticipated traffic growth while operating in a spectrum-constrained environment.

“We must operate with our minds thinking that we’re going to be spectrum-constrained. Bandwidth isn’t reducing per user. The expectations that we have is that bandwidth is just going to continue to grow. We can innovate as much as we possibly can on the physics with RF to get as much traffic density as we can on sites, but ultimately, when you dont have access to an unlimited spectrum bucket, you have got to find new ways of managing that traffic,” McElfresh reflected.

An open RAN approach, he continued, will lower the cost curve for AT&T to be able to “build more small cells or more mission-specific radiating points for particular venues or environments to manage particular types and shapes of traffic, [in] a way that we really have been unable to do with the big, main infrastructure providers.” AT&T, he explained, sees open RAN as an important enabler for AT&T to leverage its extensive fiber network combined with specialized RAN equipment in different environs and help is to “find ways to serve traffic without the need to rely upon more licensed spectrum being available in the market.”

Standardized network APIs and the implementation of rApps and xApps play in here as well. While much of the conversation around those applications in Open RAN has centered around what they can do to optimize network operations, McElfresh said only about 30% of AT&T’s focus is on such internal network applications of rApps.

The vast majority of its energy, he said, it looking at rApps “in terms of creating new capabilities or enablement for us to productize or partner with others to make the network perform in ways that we, as service providers, wouldn’t ordinarily think about,” he said, adding: “Nobody has presented at scale a canvas for third parties to actually come and code against, in a way that’s substantial enough to prove whether or not their concept or idea kind of has merit. … Our intention isn’t to keep this as an internal development track, but more importantly, open up the rApps to bringing in many third parties—as we have done in our early stage of our Standalone core development work.”

Combined with industry-wide efforts around standardization of open APIs, the opening up and modernization of AT&T’s network, then, continues its strategic position that the network is the killer app.