Your browser is not supported. Please update it.

24 May 2022

AT&T and T-Mobile come closer together on Open RAN strategies

We have noted before that the USA’s major operators have their own approach to Open RAN, quite different from that of other major MNOs – and indeed, they differ from one another in their attitudes to the emerging platform. However, while AT&T has been seen as a cheerleader for Open RAN, and T-Mobile USA as a skeptic, in fact both operators have notably similar strategies now that the technology is approaching commercialization. Their rhetoric may be different, but both are interested in Open RAN, but very cautious about timescales to deploy it in the urban macro network, or to introduce new suppliers to the 5G RAN.

AT&T, which contributed significant seed code to what became the O-RAN Alliance, is really pursuing its own RAN architecture, with considerable inhouse engineering rather than relying on industry specifications alone. It is not yet clear how far the operator will diversify its RAN supply chain, as opposed to just requiring its interfaces to be supported.

Verizon also has a heavily customized approach which ties in, more closely than AT&T’s appears to, with its advanced edge cloud developments. In this way, Verizon may end up with an architecture that looks somewhat like that of greenfield Open RAN deployers Rakuten and Dish. They are both deploying many network functions on edge nodes that are distributed out as far as the cell site in some locations, whereas AT&T – which recently transferred its own network cloud to Microsoft to run and evolve – looks more centralized in its approach to virtualized RAN.

Meanwhile, T-Mobile USA has been very cautious about vRAN and Open RAN, despite some plans to test open interfaces, and said last week that it had no plans to displace its existing vendors, Nokia and Ericsson.

It remains to be seen whether that strategy is influenced by the impact of Dish’s market entry. Dish, which also has MVNO deals with TMO and AT&T, went live with its first 5G network in Las Vegas earlier this month, has been aggressive about the cost and service agility benefits it will gain from building a cloud-native, open RAN. It claims it has actually built a distributed edge cloud, and that the 5G network is just the first use case for that cloud, which will also support many of the network-as-a-service offerings it plans to provide to wholesale customers, including its own mobile service arm.

Although AT&T and Verizon have both said they will deploy Open RAN this year, they are taking baby steps compared to their greenfield 5G challenger. At the Big 5G Event in Austin, Texas last week, AT&T did provide an update on its Open RAN activities, and its CTO, Andre Fuetsch, said the operator is now running proof of concept (PoC) tests, four years after AT&T co-founded the O-RAN Alliance together with China Mobile, Deutsche Telekom, NTT Docomo and Orange.

Like most non-greenfield operators, AT&T will deploy Open RAN first in small cells, mainly to support private networks. These are usually greenfield deployments themselves, and have less demanding coverage and capacity requirements than a public macro network. Most operators believe vRAN and Open RAN technologies will not be ready for deployment in the urban macro network until 2026 or later, and some, such as the UK’s BT, believe this will be a 6G-era development.

Some operators will deploy Open RAN in a more limited sense, implementing open interfaces between non-virtualized basebands and radio units to ensure multivendor interoperability and future-proofing. But for those interested in the full cloud-based architecture, small cells will often be the place to start, based on interfaces defined by Small Cell Forum (around functional split 6) or O-RAN Alliance (split 7.2).

Fuetsch said that the first Open RAN products would be inside buildings. He did not provide many details, but it seems likely that deployment of Open RAN small cells will coincide with AT&T’s roll-out of the midband spectrum it acquired in the C-band auction last year, as well as several smaller sales.

“Our first implementations here will be indoor and in-building,” Fuetsch said. “If you have issues with an indoor network you can always fall back to the outdoor network.” Its next stage, once it has proven the platform indoors, would be to extend Open RAN to rural outdoor networks, which are also simpler than dense, urban networks.

“We have to introduce O-RAN in incremental modules,” he added, saying that he was aiming for the “right blending” as AT&T slowly moves from traditional RAN to virtualized and open technology.

Fuetsch said last year that support for Open RAN was largely driven by AT&T’s ongoing effort to diversify its supply chain – this started with its Domain 2.0 program, which has resulted in the adoption of white box architectures in the transport network, including some new suppliers such as DriveNets.

He has highlighted the RAN Intelligent Controller (RIC) as one element where AT&T hopes to open up to multiple vendors, though has not been drawn on the potential to introduce new radio suppliers to the main network. Like Verizon, AT&T added Samsung to its 5G RAN roster alongside Nokia and Ericsson (and at Verizon, Samsung has captured significant business at Nokia’s expense in recent years). Samsung and Nokia have both committed to support Open RAN interfaces where operators require them.

Last autumn, AT&T told the FCC that it would add Open RAN equipment to its network “within the next year”, a similar timescale to Verizon’s. Both operators have talked about insisting on Open RAN from all their vendors, but neither has specified what that really means in practice. Both are major customers of Ericsson, which has not agreed to support key O-RAN specs such as the near-real time RIC or the Option 7.2x functional split, yet Verizon has even said that Ericsson will start providing it with Open RAN-compliant equipment later this year.

Fuetsch insisted that AT&T still supports the goals of Open RAN, to enable multivendor networks and a broad RAN ecosystem, but it was clear that new Open RAN suppliers, and the full cloud-based O-RAN architecture, will not feature in the macro network deployment planned for C-band spectrum from this year.

Over at TMO, there is a far more skeptical view of Open RAN and of the potential to open up the macro network supply chain significantly. At the Big 5G Event, the company’s president of technology, Neville Ray, said that expansion of its 5G footprint in the 2.5 GHz spectrum it acquired with Sprint will be heavily reliant on the huge contracts TMO has with Ericsson and Nokia. The aim is for TMO, which initially achieved broad 5G coverage by using the 600 MHz band, to extend the higher capacity of 2.5 GHz 5G to 260m users this year.

“When we looked at how far we are pushing the features and roadmaps for these vendors, the only place we could find that is with those two,” Ray explained.

Ray said the company is very interested in Open RAN but does not believe it is sufficiently robust to deploy yet. It evaluated the platform when it was planning its 2.5 GHz roll-out and will continue to test equipment as it evolves. “We love open interfaces,” said Ray. “We are pushing hard on O-RAN … we’ve got equipment we continue to test.”

However, he said a likely approach will be to ask Nokia and Ericsson to support those interfaces, as Verizon is also doing, rather than to include new suppliers. TMO does not even work with Samsung, though Ray did say that the Korean firm was one of the few with the financial resources to support the investment in Open RAN – which he believes is “a scale game, especially on the R&D”.

“I think you see AT&T and Verizon doing much the same thing as us. We are all working with our vendors to make sure the O-RAN capability is there,” he said.

Last year, Ray told an investor conference: “We just have to be patient. The ecosystem in O-RAN is going to take some time to develop. You look at the radio business today and you have three major incumbents, right? One of them Chinese. It’s a massive scale game and it’s a complex business… There’s massive R&D in this space to future-proof your networks. There’s a whole debate about who built and developed this stuff and who owns the IP.”

He said that Open RAN could not meet the capabilities TMO needed in its first stage of 5G deployment. “Those capabilities were simply not there, let alone future-proofing,” he said. But he said the firm was “supportive”, even though it would leave competitors and new entrants to drive O-RAN progress in the near term. “I’m trying to drive those pieces which I know will deliver and secure material benefit for this company, and O-RAN doesn’t necessarily help,” he said.

He concluded: “We just have to see how it develops. But we’re there. We’re watching. We’re all for open standards and capabilities, absolutely … It’s going to go through a massive journey.”