Open RAN schism and backtracking have clobbered salesOpen RAN schism and backtracking have clobbered sales

When times are hard, the weakest usually suffer the most. Data just out from Dell'Oro, one of the main analyst firms tracking radio access network (RAN) development, indicates the perennially nascent subsector for "open RAN" certainly fared much worse than the bigger market, with sales down 30% year-over-year for the first nine months of 2024. Just a few weeks ago, Stefan Pongratz, one of Dell'Oro's analysts, estimated the drop in RAN sales over this period was between 10% and 20%.

Open RAN, to the uninitiated, is basically a set of new interfaces between different RAN components to replace the older and more proprietary links used in traditional networks. The idea is to ensure an operator can match baseband technology from one supplier with a radio unit (RU) from another (or do various other combinations), instead of having to buy the two pre-integrated parts from the same vendor's system.

Unfortunately, for smaller vendors hoping open RAN would provide a boost, it is still more convenient and economical for a telco to deal with a limited number of big suppliers. Even as telcos demand compliance with open RAN specifications, they are sticking largely with "single vendor" arrangements. AT&T's $14 billion "open RAN" contract with Ericsson is probably the best example of this oxymoron.

Religious divide

A schism in the open RAN community has not been helpful and is blamed partly by Dell'Oro's Pongratz for the damage this year. Parts of the industry, led by Ericsson, previously decided the 7.2x Category B interface developed for massive MIMO, an advanced 5G technology, was not up to scratch. That interface put most uplink functionality into the baseband distributed unit (DU) to keep RUs as simple as possible. But that risked an explosion in fronthaul traffic between the DU and the RU, complained Ericsson, with the agreement of various telcos. It could lead to performance problems, they said.

Eventually, the O-RAN Alliance, the group in charge of specifications, came up with two optional "uplink performance improvement" (ULPI) alternatives. One, preferred by Ericsson, is to kick all those uplink functions back into the RU, including the equalizer, an important function that addresses interference. The second option keeps the equalizer in the DU but shifts other uplink functions into the RU.

The O-RAN Alliance was obviously conscious that RUs minus equalizers would have a tiny market opportunity if most DUs also came without this function. To avoid that scenario, it insisted that equalizers be included in all DUs regardless. The "commercial readiness" of these various ULPI products, or lack thereof, is somewhat responsible for this year's sales decline, said Dell'Oro's Pongratz in a LinkedIn post.

Fully compliant, sometimes

Since then, it has emerged that Ericsson, the world's biggest 5G vendor based outside China, is not as fully committed to open RAN as it previously suggested. It had already indicated it would not be making any RUs minus equalizers. In recent weeks, it has also confirmed that it will support open RAN for massive MIMO only when it is running software on Intel's general-purpose processors. Its purpose-built range based on its own 6672 processor – described as a groundbreaking 5G product when it was announced about a year ago – will not comply with specs.

In other words, if you want a purpose-built open RAN for massive MIMO, you won't get it from Ericsson. In areas where there is no RAN virtualization, this effectively binds AT&T to Ericsson for massive MIMO RUs and rules out its use of any third parties. And given Intel's current predicament, AT&T would have to be very brave to embark on a widespread deployment of a virtual RAN. Despite the best efforts of AMD, a rival, and other chipmakers using the blueprints of UK-based Arm, Intel remains pretty much the only virtual RAN option. Much has been made of AT&T's plans to use Fujitsu and now Mavenir as RU suppliers. But unless AT&T finds another DU partner, they won't be doing the sexy stuff.

Nokia's Tommi Uitto, who heads up the Finnish vendor's mobile networks business group, had smelt something iffy when he spoke with Light Reading at the start of the year. "If you have baseband in the field, like one of my big competitors, that doesn't support the equalizer, they cannot connect the baseband that doesn't have the equalizer to an RU that doesn't have the equalizer because there is an equalizer nowhere," he said. "And that is why a big competitor has a problem with the installed base. You have to change to a new baseband or do cloud RAN. You can't use the existing baseband."

More recently, Samsung has distanced itself from the Ericsson approach. "We don't really see the point in that distinction," said Alok Shah, the vice president of strategy, business development and marketing for Samsung Electronics America, insisting both purpose-built and virtual RAN products from Samsung are fully compliant. "We're O-RAN-capable across the board."

Even with the new interfaces, however, Nokia concedes that pairing vendors in massive MIMO remains difficult. Its own open RAN contract with Deutsche Telekom for 3,000 sites in Germany involves providing both DUs and RUs in massive MIMO zones. Elsewhere, the plan is to pair its DUs with RUs from Fujitsu. At AT&T, where there has been recent talk of small cells but little certainty, Fujitsu and Mavenir may be lucky to get even so much as that.