Years ago, open radio access network (RAN) fanatics would talk about Ericsson and Nokia in the scathing language Greta Thunberg reserves for oil companies. The Nordic equipment vendors, destined to lose business if open RAN took off, were portrayed as tyrants determined to block change and keep telcos hooked on their proprietary goods. Fast forward to 2024 and both now profess support for the concept and its specifications. 

Confronted with the inevitability of it all, they have finally backed down like an unpopular government ceding control. Or have they?

Figures from multiple sources show Ericsson and Nokia have grown more powerful as they have sought accommodation with open RAN. Fresh from securing a $14 billion "open RAN" deal with AT&T in the US, Ericsson was boasting market share gains in Europe, India and North America in the earnings report it published last month. Outside China, its share of the RAN market grew from 33% in 2017 to 39% in 2022, it has claimed.

Despite some high-profile setbacks in the US, including the loss of AT&T business to Ericsson, Nokia has also been in the ascendant. In the last two years, according to CEO Pekka Lundmark, its 5G market share outside China has grown from 23% to 29%. No doubt, Ericsson and Nokia have profited at Huawei's expense as western governments have curbed the activities of the security-threatening Chinese vendor. Yet the smaller vendors that were supposed to be the main beneficiaries of open RAN have barely advanced.

A report published last year by Omdia, a Light Reading sister company, provides evidence. It showed the revenue market share of the top five RAN vendors – Huawei, Ericsson, Nokia, ZTE and Samsung – declined by just half a percentage point between 2021 and 2022, to 94.6%. Dell'Oro, another analyst firm, reckoned 2% of revenues in a shrinking market were split between 30 companies for the first nine months of 2023.

No multivendor to be seen here

Despite the opening of network interfaces by major kit suppliers, including Ericsson and Nokia, big telcos have continued to sign deals with a single vendor for both compute (or baseband) and radio. The original purpose of open RAN was to smash the proprietary link between these two main components of the RAN, allowing an operator to buy each one from a different vendor. It is not happening at scale and won't for years.

Recognition of that comes from Stefan Pongratz, a Dell'Oro analyst. By 2028, open RAN will represent between 20% and 30% of the entire RAN market globally, he said in a recent LinkedIn post. Yet the "lion's share" of these open RAN deployments will fall under the "single vendor" category. Four years from now, the "multivendor" form of open RAN is expected to account for just 10% of the overall RAN market in the best case, and for as little as 5% in the worst.

A common response among those still optimistic about open RAN as a market gamechanger is to argue that compatibility with specifications is merely a first step, the flame that will light a glorious firework display of multivendor contracts. But specifications were arguably never the main problem, and the widespread availability of compatible products will conceivably have little impact on the nature of the commercial relationships between vendors and telcos. Ericsson and Nokia feasibly calculated that open RAN was not the business-decimating threat it was first cracked up to be.

What no specification will ever do is allow an operator to fit together two separate vendors like a child building Lego. Even within the same vendor's organization, the job of integrating compute and radio products is immense, and someone needs to pay for it. "When I introduce a new radio, I have a lot of integration work to do with baseband and lots of testing to do," said Mark Atkinson, Nokia's head of RAN, during an interview with Light Reading last year.

Pairing vendors, then, means finding someone to do this systems integration job, and paying for it. And concerns about additional cost and complexity are likely to explain why AT&T appears to have left that all-important role in the hands of Ericsson, and why most of its "open RAN" site equipment will come from the Swedish vendor. Fujitsu, the only other radio vendor named in the deal, had seemingly worked to integrate its products with Ericsson's before an open RAN specification even existed.

Staggering numbers

AT&T was certainly under no illusions about the scale of the systems integration challenge weeks before the Ericsson deal was announced. "Those numbers that are spent on R&D by the incumbents are staggering," said Robert Soni, AT&T's vice president of RAN technology, at the FYUZ event in Madrid last October. In a "closed ecosystem world," he estimated that "one in three dollars is spent on integration and validation." The remarks came shortly after Soni reckoned Ericsson was investing about $3 billion annually in R&D for RAN technology. "Disruptors are a fraction of that," he said.

The alternative to recruiting a systems integrator that does not develop RAN products, and therefore represents a discrete cost item, is for operators to do the work themselves. Vodafone, notably, is adding thousands of software engineers to avoid reliance on external systems integrators. During an update last week, Scott Petty, Vodafone's chief technology officer, revealed it has reached 7,500 of a targeted 9,000 hires since the plan was announced in late 2021. Today, he said, Vodafone employs 14,000 software engineers in total.

But none of this is cheap. Nor does Petty think generative artificial intelligence, with its code-writing abilities, will help to minimize staff-related costs. "I don't see it as an efficiency gain at all," he said. "It helps a lot in documentation, but the productivity gains and 40% you hear of are wildly optimistic in terms of what we've seen. It's a good tool to take away some of the drudgery of software development."

AT&T's Soni, moreover, evidently doubted back in October that any telco could succeed against the "closed ecosystem world" on its own. "The only opportunity for open RAN to catch up to that ecosystem – and provide performance and feature parity but also leapfrog in terms of capabilities – is for us to pool," he said. Since then, of course, Ericsson looks to have assumed the exact role in AT&T's open RAN deployment it would play in a closed RAN.

The establishment of test and validation centers by the likes of the Telecom Infra Project, which organizes the annual FYUZ event, should certainly help with the pooling of expertise and resources, and with knowledge dissemination. But "single vendor" open RAN is moving faster. An operator buying baseband and radio from a single supplier probably gets a better per-unit price than a telco signing two separate contracts, just as customers have seen the cost benefits of taking phone, broadband and TV services in one deal.

The drawback, as telcos know, is commercial lock-in. A telco might be able to negotiate the switch to another vendor's radios while it retains the original vendor's software. But removing hardware before it has fully depreciated can be expensive. The more "single vendor" deals the big vendors can sign now, the less they will have to worry about losing market share for years.