If the Open RAN community ends 2022 on a rather disappointed note, it will be partly because the expectations set for the platform were always over-ambitious. Expecting operators to deploy a brand new architecture at rapid pace and broad scale, in their macro networks, would always have been very optimistic, even if that architecture would not also be a) their first foray into virtualized RAN; b) based on interfaces that were not, except in the case of Open Fronthaul, completed and hardened; and c) needing to coexist with ‘legacy’ networks that had sometimes only been rolled out 2-3 years before. To add unfamiliar and small-scale vendors into the mix too just added to the perceived risk.
This is not to say that all these challenges will not be addressed, given the large amount of industry collaboration and innovation that is ongoing, but they will take several years to be addressed to the extent that operators will feel safe to entrust their macro networks to the new platforms and vendors. vRAN designs and cloud infrastructure that can effectively support Massive MIMO; the cost and risk of integrating multivendor systems; orchestration of new and old elements – all these are thorny issues.
No surprise, then, that most estimates show that the established RAN vendors have scarcely been touched by Open RAN so far, and if they are threatened in 2023, it will be mainly in small cells and enterprise networks, which are developing their own parallel ecosystem anyway.
Even many of the significant Open RAN macro commitments that have been announced in 2022 have been focused on relatively low-traffic cells in rural areas, and have been supplied by established, tier 1 equipment vendors that have decided to embrace the Open RAN cause in the name of increasing their market share. Rakuten with NEC, Dish and NTT Docomo with Samsung – these deals may improve these Asian vendors’ ability to challenge Ericsson, Nokia and Huawei, and so provide more choice for operators, but they do not usher in the world of small suppliers and multivendor RANs that many had envisaged.
One new report, from Dell’Oro research group, found that the top five RAN vendors – Ericsson, Huawei, Nokia, ZTE and Samsung – have collectively lost only one percentage point of RAN market share in the past year. Most of the change has been in how they split their 98% among themselves, with Huawei’s share under pressure in US-allied countries but strengthening in China.
Dell’Oro does think that growth in Open RAN has been faster than expected in 2022, and that could deliver it a share of the RAN market of between 6% and 10% by the end of 2023 (though much of that could still be in the hands of Nokia and Samsung). And it thinks the growth rate will decelerate next year amid recessionary factors.
“Open RAN continues to move forward and accelerate at a faster pace than expected,” said the company in its latest report, and it calculates that third quarter Open RAN sales doubled year-on-year. As well as Nokia and Samsung, NEC, Fujitsu and Mavenir were the leading vendors of Open RAN products in the first nine months of 2022, and all of these have notably moved away from specializing in one part of the RAN, to providing end-to-end solutions, directly or through partners. NEC even includes the xHaul in its end-to-end definition, via its own microwave division and its alliance with Juniper.
Samsung, so far, looks to be the big winner in Open RAN, using support for open interfaces and vRAN to consolidate the expansion it has made into 5G anyway, seeded by its major deals with Verizon and AT&T in the USA. It is supplying Vodafone with equipment for the firm’s UK replacement of Huawei systems, which will eventually reach 2,500 sites. It is not clear how many of those sites will be Samsung-supplied, but only one Open RAN macrosite is live within Vodafone UK to date and the bigger push is likely to come from late 2023.
Dell’Oro and Rethink findings both point to declining interest in truly multivendor deployments like Dish Network’s, because operators are reacting against the integration and management complexity, and the difficulty of managing many suppliers. Even Dish, which initially chose Fujitsu and MTI of Taiwan to supply its Open RAN radio units, then added Samsung in May this year, and implied in investor meetings that the Korean vendor’s scale had saved Dish from delays that were threatening to land it with regulatory sanctions.
Dish Wireless’s chief commercial officer, Stephen Bye, said at the time that the addition of more TDD bands to the firm’s spectrum rainbow was one factor behind the selection of Samsung, since the Korean firm was a pioneer in Massive MIMO solutions for the 5G TDD midbands. But behind those words, it was clear that Samsung could supply radio units at a faster pace than Dish’s smaller suppliers, helping it to populate the new spectrum and therefore meet its coverage targets.
Bye was insistent that these were reasons for the Samsung RAN deal, not any problems with existing suppliers or with integrating a multivendor network. But in August, Dish chairman Charlie Ergen said the addition of Samsung to the supplier roster had made a big difference to the pace of roll-out. It gave Dish a second source for Open RAN radios, in addition to Fujitsu, which helps safeguard supply and drive scale. And Ergen said Samsung has significant expertise in optimizing the network and devices to support Voice over NR, as well as Band 70, the idiosyncratic spectrum band that lies at the heart of Dish’s multiband roll-out.
In an interview with LightReading, chip provider Marvell’s senior director of product marketing, Joel Brand, summed up the thoughts of many RAN stakeholders, saying: “On the radio side, you have Ericsson, Nokia, Samsung, Fujitsu and NEC outside China. That’s five vendors – two strong ones and three to kind of keep them honest – and that is how these industries work. It is not particularly unique to wireless … We need to keep in check what we are trying to do, and this is not about building an ecosystem of 100 players.” Instead, Brand thinks Open RAN’s value lies in providing “open interfaces that allow smaller companies to plug in their stuff and innovate”.
Some operators, such as the UK’s BT, have certainly cautioned that they do not wish to work with more than two or three major vendors, though they would welcome more choice of who those vendors might be. But the dream of swappable, mix-and-match networks with a huge choice of suppliers of different elements always comes up against the integration challenge – at least until a really plug-and-play platform is devised, which seems highly unlikely outside the small cell world.
Two of Open RAN’s biggest champions, Deutsche Telekom and Vodafone, have both spoken recently about the need to address integration issues. DT had initially said, when it built its ‘O-RAN Town’ testbed in Neubrandenburg, that it hoped to be deploying citywide Open RANs across Germany in 2023.
Instead, sources say that DT has put its request for quotes (RFQ) on hold because systems are not yet robust enough to consider for large-scale production networks, and integration challenges must be thought through.
“Regarding open RAN, Deutsche Telekom is still in the process of qualifying vendors in preparation for a commercial roll-out from 2023/24,” said a spokesperson. In O-RAN Town, the trial vRAN runs on Intel-based servers from Dell and Supermicro with Mavenir software and radios from Fujitsu and NEC.
Even Open RAN vendors are acknowledging operator caution. During an investor presentation in September, the president of NEC’s network services business, Atsuo Kawamura, said: “Compared to greenfield operators like Rakuten Mobile, brownfield companies tend to be cautious to Open RAN deployment. They conducted Open RAN tests last year and with that result they have gone somewhat cautious.”
Speaking at the NGMN conference in Paris in September, Arash Ashouriha, DT’s deputy CTO, said systems integration was one of the biggest challenges to build Open RAN infrastructure; while Yago Tenorio, Vodafone’s network architecture director and the chairman of open networks initiative Telecom Infra Project, said the idea of plug-and-play Open RAN was “totally unrealistic”. Tenorio said the prospect of lock-in by an integrator rather than a vendor was a real one, saying: “If there is nothing else but a systems integrator, the risk is you go with Tech Mahindra, they have all the knowhow, and you are doing Open RAN but depending on them.”
Some operators will avoid that risk by doing their own integration, if they can afford the time and resource, but many are hanging their hopes on more standardized platforms and testing processes. Francisco Martín Pignatelli, group head of Open RAN at Vodafone, told the recent FYUZ conference – a joint effort between TIP, O-RAN Alliance and Meta – that: “Collaboration is the key word and the key challenge from our side. We need to start promoting these types of collaborations so that we don’t repeat tests or build multiple labs… And it’s something that should happen between operators, but also with the supplier ecosystem.”