Ericsson defends 'openness' of $14B AT&T deal

"This is the biggest contract in Ericsson's history," said Fredrik Jejdling, the seemingly unflappable head of Ericsson networks, the business group responsible for about two-thirds of the Swedish vendor's roughly $25 billion in sales last year. He was speaking, of course, about the recently struck $14 billion deal with AT&T, an arrangement that will mean replacing Finnish rival Nokia across the one-third of the US operator's mobile network Ericsson doesn't already serve. It's not only Ericsson's biggest contract ever; it's also supposedly the first "open" radio access network (RAN) contract Ericsson has signed.

To opponents, that is a gross non sequitur. Open RAN was initially conceived by telcos to end the tyranny of "lock-in" by three dominant RAN vendors – Ericsson, Huawei and Nokia. With its new interfaces, an operator should, at any given mobile site, be able to host a party of compatible suppliers, each a specialist in its field, rather than being stuck in a single vendor system. Adding virtualization, a cousin of the concept, would bring the advantage of cloud economics. Yet this deal, apparently worth almost $3 billion a year, makes AT&T even more dependent on the West's biggest 5G vendor, a company still portrayed as an enemy of open RAN just a few months ago.

Related:What's the Story? The reverberating revenue ramifications of the AT&T and Ericsson open RAN deal

Ericsson, though, seemed to overcome its antipathy after the O-RAN Alliance, the telco-led group behind specifications, signed off on an important fix. Thanks to an agreement reached in Japan last June, Ericsson can now develop compatible products for massive MIMO, an antenna-rich 5G technology, and keep some of the important uplink functions in the radio. Under the original specification, those had to be in the baseband unit, which Ericsson saw as a problem.

"We think now we can build radios and baseband performing in a similar way, at the same level, as we can do with our current fronthaul specification. And that was important because we didn't, and that was one of the hurdles," Jejdling told Light Reading. "We have no intent to be in any other fronthaul or northbound interface."

From open RAN skeptic to advocate

Ericsson will not, however, be cooking up all flavors of open RAN. Under a compromise brokered by the O-RAN Alliance, companies unhappy about shifting uplink functions into the radio will be able to rely on baseband capability instead. Ericsson has ruled out making simpler massive MIMO radios – which tells observers something about the nature of AT&T's plans.

That's because massive MIMO will figure in the broad range of products Ericsson sells to AT&T. "In this deal, we offer radios, we offer baseband, we offer cloud RAN, we offer Ericsson's intelligent automation platform – a big subset of our portfolio," said Jejdling, citing the "latest-generation Ericsson massive MIMO" as one component. Indeed, it is such a big subset that opportunities for other vendors appear limited. Pekka Lundmark, Nokia's CEO, has notably described it as a "one vendor" deal.

What seems increasingly evident is that any role for others will probably not extend beyond the provision of radios at some AT&T sites. "There are alternative vendors in there on the radio side, as well, for example, and that comes naturally with an open RAN deal," said Jejdling when asked how this can be justified as an open RAN contract.

Right now, Fujitsu is the only other RAN company that AT&T has identified as a vendor, and the rough percentage of sites it might serve remains unclear. "That I can't answer because that depends on what type of equipment AT&T would ultimately buy into an open RAN network," said Jejdling. Marrying its radios with Ericsson's RAN software, at least, should not be a problem. Ericsson and Fujitsu have had a 5G partnership since 2018.

This would not strictly qualify the rollout as an open RAN, but merely prove that two vendors have achieved compatibility. Unlike Nokia, Ericsson has publicized no details of interoperability tests with any other vendors. During Open RAN North America, a December event organized by Informa (Light Reading's parent), Mavenir, Nokia and Samsung all indicated they had not had the opportunity to test open RAN interoperability with Ericsson in any forum. Nor did the O-RAN Alliance have anything about Ericsson interoperability tests on record, it told Light Reading in December.

Nevertheless, all the products supplied by the Swedish kitmaker will be fully compliant with O-RAN Alliance specifications, according to Jejdling. "That comes naturally – if it weren't that, it wouldn't be an open RAN deal," he said. "That's what AT&T was interested in and that's what we wanted to provide to them."

Ideally, for AT&T, this should mean it can easily introduce other radio suppliers alongside Ericsson and Fujitsu as the rollout progresses. Yet swapping out either of the named vendors seems unlikely. AT&T is already replacing "a significant amount of Nokia equipment, which they have not yet depreciated," Lundmark told Light Reading last month. Ditching Ericsson or Fujitsu radios before the end of their natural lifespan would only drive costs up further.

The opportunity for Intel and Dell, as hardware vendors integral to the cloud RAN concept, might also be relatively small. "There is purpose-built RAN compute and there is a component of cloud RAN solutions in there as well," said Jejdling. The message from Ericsson is that Intel's general-purpose processors won't ever close the performance gap with its own application-specific integrated circuits. When customers choose cloud RAN, it is because their priorities are programmability and resource sharing, it has said. AT&T's decisions about the rollout, then, come with trade-offs.

Margin pressure

The contract won't provide any "meaningful" uplift to Ericsson's revenues until the second half of this year, according to the company's latest financial update. And while it will boost Ericsson's market share in North America, it has not led Ericsson to raise its long-term EBITA (that's earnings before interest, tax and amortization) margin target of between 15% and 18%. To help get there, moreover, Ericsson last year cut about 9,000 jobs (with contractors included), realizing annual run-rate savings of around 12 billion Swedish kronor (US$1.3 billion). Further cuts are now promised as CEO Börje Ekholm strives to "pare back some of the investment areas we've had."

The analyst concern will be that Ericsson's juicy deal with AT&T puts additional pressure on margins, as RAN contracts often do in the early stages of rollout. Jejdling declined to comment on the profitability implications but acknowledged Ericsson had won the contract against "tough competition." Unlike most other deals, this one also means replacing equipment that has not yet fully depreciated, judging by Lundmark's comments. There is some industry suspicion that Ericsson will bear the cost.

It will not, though, be hiring large numbers of North American employees to do the work. This is, at least, the assertion of Ekholm, who insisted the modern-day Ericsson has grown more heavily reliant on third parties than in-house service engineers. It came after Joseph Zhou, an analyst at Barclays, noted on this week's quarterly earnings call with Ericsson executives that headcount reduction in North America last year was "huge." AT&T, meanwhile, has denied by email that any of its staff will be transferred to Ericsson as part of the contract.

More details will inevitably trickle out this year as work starts. But provided Ericsson makes good on its promise to be fully compliant with O-RAN Alliance specs, nobody can accuse the dealmakers of lying. Compliance alone would define this as an open RAN deal, even if Ericsson does provide most of the parts, and even if those parts are purpose-built. In that scenario, this could all be very good for Ericsson, further boosting its position in 5G. What it does for vendor plurality is another matter altogether.