Here’s the money, now open up a bit

Not wise before the event, and not very wise after it either. But while AT&T's rug pull floored Nokia, there is reason in its decision to pay Ericsson well to drive a more Open RAN strategy.

What happened there, then?

AT&T is giving Ericsson a five year run at its RAN, in a deal worth up to $14 billion. It said that the Ericsson investment would follow Open RAN principles, name checking other vendors involved in the RAN refresh as Fujitsu, Dell and Intel. At this time, as everyone has noted, there’s no role for Nokia, previously a key AT&T RAN supplier and also historically more committed to Open RAN than Ericsson. Nokia itself said that this meant revenue from AT&T in Mobile Networks will decrease over the next 2-3 years, adding that AT&T accounted for 5-8% of Mobile Networks net sales year-to-date in 2023.

In the two days since the announcement, TMN has spoken to operators, vendors and other sources to work out why AT&T moved the way it did, and why Nokia lost out in the deal that targets an upcoming RAN refresh at AT&T?

AT&T judged this was the moment to extract a clear commitment and action from Ericsson to open up interfaces in the RAN

What happened with Nokia? Can its loss really be about fan cooling on its RUs, as analyst Earl Lum posited in a piece that scooped the industry on AT&T’s intention to cut Nokia from its next round of RAN investment? Nokia sources have been unable to go on record but two have pointed out to TMN that Nokia does have RUs without fan cooling, and has sold product into plenty of hot climate markets. Nor do they concede that Ericsson won because its Cloud RAN approach, which runs its vRAN DU-CU software on Intel based servers (it has announced integrations with Dell and HPE edge servers), performs better than Nokia’s AnyRAN approach, which runs L1 processing on a Marvell SoC and can also plug into a variety of hardware.

Instead, Nokia voices have heavily hinted that this has been about Ericsson being willing to go down to a price level that won them the deal. Publicly too, Nokia put out an official statement – that AT&T will have endorsed – in which AT&T says that Nokia’s products are still very good. The clear implication from Nokia is that this wasn’t purely, or even mostly, a technical choice – and AT&T also seems willing to go along with that. (One Ericsson voice put it down to the quality of the sales relationship with the operator, but that’s not something that anyone can independently verify.)

But whilst a price commitment was no doubt key, just as key seems to have been that AT&T judged this was the moment to extract a clear commitment and action from Ericsson to open up interfaces in the RAN, enabling AT&T to move to a RAN that has the efficiencies of being automated, operated and managed as a single system, but that harvests technology from multiple vendors.

But if this is the goal why choose Ericsson as the vehicle, where Ericsson has been regarded as a foot dragger in Open RAN, rather than the more active Nokia? Well, perhaps precisely for this reason. If this is really about AT&T using current vulnerability to leverage a commitment to greater open-ness in the long term, why not use that opportunity to address the behaviour of the most difficult child?

AT&T’s reasoning

AT&T’s answers to a range of questions added little context beyond its public announcement.

Sepcifically it ingnored a question asking if Ericsson would act as systems integrator for the new Open RAN builds. The issue of who will build, manage and operate multi-vendor networks, taking responsibility for system performance, is a crucial one, and will affect the economics of this deal.

But at an investor conference AT&T CEO John Stankey did add more detail on the operators’ thinking. First off, he said that people shouldn’t get the announcement “out of context”. He said that while the $14 billion contract over five years is “material” for the vendors, for AT&T it’s not actually that big a deal. Its annual capex is $24 billion (it’ll be a little lower soon), so a deal that maxes out at under £3 billion per year is only a portion of its overall network investment – and also only a portion of its RAN investment.

But what Stankey also pointed out is that the operator knew that the vendors were in a weak moment, and judged it would be an opportune time for the carrier to twist their arms to make commitments to open up their network and (left unspoken but surely relevant) to make concessions on price.

In his words, “With the slowdown in the vendor markets, we were able to step back and say, ‘What can we do to get an opportunistic agreement where we can drive vendors into a position to move more aggressively on O-RAN to position us long term?'”

Stankey said that in this instance “long term” means “whenever the next big investment cycle comes.”  That could be when the operator gets more spectrum or if and when 6G comes around. (That approach, by the way, seems at odds with a view that Ericsson is about to rip out the 35% of the AT&T RAN that is currently Nokia’s and replace it with its own product. It seems more likely at this point to be a slower, more phased approach.)

Whilst Sankey does talk about being able to reduce “unit cost” as a result of the Ericsson deal, he added that AT&T has been able to “take advantage” of a “lull” in the suply base to move to a more open platform in its RAN. And let’s remember that AT&T has, since the formation of the O-RAN Alliance, been one of the operators most committed to a disaggregated RAN. Back in 2019 its then-champion Andre Feutsch (now EVP and CTO Network Services at AT&T) said that an open, disagreggated RAN was not “rocket science” and that those that doubted O-RAN should “follow the money.”

So let’s follow the money.

Here’s Stankey:

“We stepped back, looked at it and said, this was an opportunity for us to do that [move to a more open platform in the RAN]. And we had two very good suppliers. They both did good work for us. They both had really good equipment. We stepped back and said, how can we get to the most modernised network that gets the most amount of traffic across open — potentially open — interfaces. And it was this path that we chose with Ericsson.”

That’s an interesting self-correction there. From “open” to “potentially open” interfaces. Stankey is acknowledging that this contract isn’t about AT&T moving day one to Open RAN, it’s about leveraging a moment of weakness in its suppliers to extract a commitment to open interfaces that in time, AT&T will be in a position to exploit to introduce new vendor innovation. In return, Ericsson gets a five year commitment to one of the biggest RAN deals in its history.

And indeed in the official announcement AT&T made it plain that the initial work Ericsson does will see mostly an all-Ericsson RAN, with perhaps some attached RUs from Fujitsu.

Fujitsu, by the way, has confirmed to TMN that it will be delivering RUs to the operator. “In order to support the acceleration of Open RAN in AT&T’s network, Fujitsu will initially be delivering a series of O-RAN compliant radios to address their entire spectrum holding. More details will be provided in the near future.” 

As the open platform matures, then there could be room for other vendors to take share – and remember that $14 billion over 5 years is not AT&T’s entire RAN budget.

Stankey: “By 2026, we’ll have about 70% of our traffic across infrastructure that effectively will have interfaces on it that could be opened up for other supply. And that allows us then to think long term as we get the decisions on expanding the network and pushing it out further, how we choose to use a vendor base on that.

“It’s entirely possible. Nokia could be one of those suppliers of that more diverse vendor base that we ultimately start working toward. We’d obviously like to see multiple players there. We think that’s how we manage cost in the network, but also a degree of innovation on the packaging – and that’s going to become more important as networks become more distributed.”

An AT&T statement to TMN said that the operator would have Fujitsu RU and Ericsson baseband sites live in 2024. Then the collaboration with Ericsson paves the way for AT&T to mix and match vendors using open RAN interfaces. “Then in the coming months/years, we’ll work to integrate equipment into its network from a variety of suppliers.”

Remember too, that the 70% traffic ambition does not equate to 70% of the network – you can carry 70% of network traffic through a much smaller percentage of high volume sites.

Conclusion

The irony for many is that it is Ericsson that AT&T has tasked to drive towards an open platform that allows it to plug in other vendors.

Alok Shah, VP of strategy, business development and marketing in the networks business of Samsung Electronics America, told TMN, “It is quite a novel approach to kind of give full market share to one vendor and ask them to to introduce others, to open things up. I think the incentive structure is going to be a bit twisted in that approach and ultimately it will be it’ll be interesting a couple of years from now to see if the company’s achieved what they wanted to on the timeline that they wanted to.”

But looked at another way, choosing Ericsson makes good sense for AT&T, it gets the short term pricing benefit of the Ericsson deal whilst also holding the vendor to a commitment to open up.

Even so, we’re still back with why it went with Ericsson at this point and not Nokia, or both. And for that it looks like we are looking at business relationships, pricing and where AT&T saw maxiumum value in extracting alignment with Open RAN principles –  as damaging as it is and frustrating as it must be for Nokia.

So the biggest beneficiary of the largest brownfield “Open RAN” contract to date is Ericsson. Ericsson is committed to Open RAN if you believe that its Cloud RAN plus support for the ULPI spec in O-RAN fronthaul qualifies. And plenty do believe that, such as Telefonica and probably Vodafone too. Look for more such Ericsson “Open RAN” announcements in 2024.

The piece earlier this year that said that Open RAN was at a cliff edge, as the major vendors position themselves to be Open RAN providers, is looking pretty accurate. Including the bit about the fall.

As this piece from Heavy Reading’s Gabriel Brown says, operator benefits from Open RAN will be achieved by striking a balance between integrated and open, old and new. AT&T’s move with Ericsson looks to have put the thumb largely in the single vendor side of the scales. But over time, the operator said it is willing to load up the other side of the scale with more vendors. Those vendors will have to take AT&T at its word, just as AT&T will have to keep Ericsson to its own commitments.