Unthinkable as it might now seem, Rakuten was poised to sign a contract with Huawei for the construction of Japan's fourth mobile network when Tareq Amin decided against it. "I wanted to re-imagine how future networks should be built," the chief technology officer of Rakuten Mobile told UK politicians last week.
Cancelling arrangements with a Chinese vendor now on many countries' blacklists, Amin subsequently embarked on the world's most ambitious network project. His inspiration was the standardized kit and cloud-hosted software he had witnessed on a previous visit to one of Facebook's data centers. And central to his plans was open RAN, an in-vogue technology that promised alternatives to Huawei, Ericsson and Nokia, the dominant providers of traditional radio access networks.
Nokia has figured in Rakuten's rollout, supplying radio units for the 4G network, but the spotlight has shifted to others. On the open RAN side, the most prominent vendors are NEC, a Japanese rival to Nokia, and Altiostar, a US software startup. Today, it is majority owned by Rakuten in a sign of the Japanese firm's desire to control and shape technology development.
Rakuten's big opportunity is in selling these technologies, packaged up as the Rakuten Communications Platform (or RCP), to overseas networks. Authorities have made a willing accomplice. "The Japanese government has provided subsidies to Rakuten to continue to build what is called RCP," Amin told UK officials.
Japan is not the only country that sees open RAN as an opportunity to nurture government-preferred technologies or even homegrown providers. In the UK, which is banning Huawei, some politicians have already backed open RAN, an immature and largely unproven concept, as the winning technology formula. "Open RAN is obviously the ultimate approach on the standards front," said Aaron Bell, a member of the governing Conservative Party, during a parliamentary enquiry that featured Amin last week.
Similar forces are at work elsewhere. Timotheus Höttges, the boss of German incumbent Deutsche Telekom, has pressured his government to mandate the use of open RAN. Attuned to their government's protectionist instincts, Indian operators Bharti Airtel and Reliance Jio are prioritizing made-in-India technologies. Open RAN is reportedly high on Airtel's list.
A US lobbying group has also popped up to persuade friendly governments of open RAN's appeal. Led by Diane Rinaldo, a former civil servant and advisor to the US government, the Open RAN Policy Coalition features not a single Chinese member, despite the involvement of numerous Chinese companies in the O-RAN Alliance, the group behind open RAN specifications.
Inflating the bill
This lurch toward protectionism and state intervention reflects modern geopolitics. The tech clash between China and the US has claimed Huawei as a high-profile victim. Its probable demise in Europe, India and other pro-US markets creates a big vacuum. Filling that with Ericsson and Nokia would leave service providers, and the economies they support, at the mercy of a Nordic duopoly. That creates a dilemma for authorities. "The chances of creating another Nokia in the UK are close to zero, and I'm not sure it would be the right thing anyway," said Ian Livingston, the head of a UK taskforce set up to explore alternatives, and a former CEO of BT.
Cultivating an open RAN ecosystem strikes many politicians and operators as a more viable option. The "softwareization" of networks extends a possible role to startups with a tiny fraction of Ericsson's resources. Their technology could be used with general-purpose hardware, driving down expenditure. Amin says Rakuten's network cost 45% less to build than a traditional one. Thanks to new O-RAN interfaces, operators would have the freedom to use components from different suppliers at the same site – something they complain is not possible today.
This vision of openness is clearly at odds with a pick-the-winners approach. Nor does protectionism typically result in lower costs and greater efficiency. Open RAN might conceivably support a larger number of equipment vendors than old-school technology. It would not allow every country to insist on homegrown technologies without inflating the service provider bill.
It is worth considering today's network economics, and how the telecom sector arrived at a point where three giant vendors account for most of the market. Turn the clock back far enough, and the equipment sector used to be a crowded community, featuring the likes of Alcatel, Lucent, Nortel and others that have since disappeared. The UK alone once had GEC, Plessey and STC, names that will be unfamiliar to many industry employees now.
Successive rounds of consolidation have left operators with far less choice in 2020. Yet Ericsson and Nokia hardly fit the profile of oligopolists fleecing their customers. Ericsson, considered the healthier of the two, was loss-making as recently as 2017. After clawing its way back to profitability by selling assets, shrinking its workforce and doubling down on radio, it is confident it will achieve an operating margin of at least 10% this year. Nokia's came in at only 6.6% in its last fiscal quarter, and a disappointing outlook contributed to a sharp drop in its share price when results were published. Any notion their products are overpriced does not stand up to scrutiny.
Contrast those figures with customer equivalents. Deutsche Telekom, Europe's biggest operator, made an operating margin of 11.4% for the June-ending quarter. Telecom Italia managed 9.3% for the first six months of the year, while the UK's BT last week reported a 9% margin for its own fiscal first half. While Vodafone does not disclose profitability on a quarterly basis, its operating margin last year was 9.1%. All four of these operators have recently done better than their biggest European suppliers.
They have benefited, too, from unparalleled economies of scale when buying radio equipment. Ericsson and Nokia combined employ nearly 200,000 people, maintain facilities across the planet, spend almost $9 billion a year on research and development and cater to a global audience. All that translates into lower per-unit costs than a balkanized market would enjoy.
A new form of lock-in?
How, then, can Amin boast a 45% saving with systems that only his network is using? Possibly because, as a greenfield operator, he has been able to simplify his architecture and rely heavily on automation. In any case, the cost verdict has not yet been served on Rakuten, which is still recording losses on network deployment fees.
Also unresolved is whether Rakuten has sacrificed quality in its open RAN experiment. Amin insists not and says metrics showing high levels of data usage per customer are the proof. That is easier to manage when customer numbers are so low, skeptics retort. Rakuten had only 1 million "applications" for service at the end of June and has not disclosed take-up since then. Its 5G network is also slower than rivals, although speed improvements are promised (and may be announced this month).
Regardless, many brownfield operators are unpersuaded Rakuten is a blueprint. Howard Watson, the chief technology officer of BT, does not expect to use open RAN technology until late 2026, at the earliest. He has picked Ericsson and Nokia for his company's 5G upgrade. Vodafone is more gung-ho, revealing plans this week to substitute open RAN for Huawei at around 2,600 UK sites. But this would still be a relatively small fraction of the total – and concentrated in less demanding rural areas. No other European service provider has gone even that far.
Besides driving up costs, picking winners could undermine the whole principle of openness. Amin presents RCP as an open platform that will offer choices between different technologies, but Rakuten's ownership of Altiostar gives it a vested interest in a specific open RAN developer. Parallel Wireless, a rival to Altiostar, is dubious. "The challenge for RCP will be in adhering to open principles while simultaneously continuing to pour resources … into Altiostar," said Steve Papa, the CEO of Parallel Wireless, in emailed remarks.
RCP and a related tie-up between Rakuten and Spain's Telefónica, which owns a minority stake in Altiostar, are also a concern for Roberto Kompany, an analyst with Analysys Mason. Under the partnership, the two operators are to pool their procurement efforts. While that is good news for their own preferred suppliers, "it may ultimately limit other vendors' access to the market, thereby reducing the level of true openness," writes Kompany in a research note. Meanwhile, operators choosing the RCP route "may be limited to the vendors and features on Rakuten's roadmap," he says.
This could also be true of platforms supported by the US public cloud providers. Amazon, Google and Microsoft have been steadily advancing into the networks market. Much like Rakuten, Microsoft has been acquisitive, snapping up Affirmed Networks and Metaswitch, two developers of core network software, earlier this year. Adding open RAN, through either takeover or partnership, would give it a full stack of network software, according to a Microsoft presentation.
Papa, who thinks this development is feasible, envisages two scenarios. "Either each cloud has its own captive stack or there will be a cloud with a captive stack and others with a more open stack," he says. Suggestions there could be a "captive stack" will unsettle anyone who worries about "lock-in," or being tied to a foreign provider's systems. After all, escaping the clutches of Ericsson, Huawei and Nokia is why many operators want open RAN.
Protectionism could be wasteful. Quizzed by UK politicians last week, Livingston effectively warned authorities not to favor specific national suppliers or build a strategy entirely around open RAN. "No supplier big or small is going to create a world-beating product just for the UK, so having a consistency of approach across a number of countries – where they know they can scale up … that is a core part of it," he said. "What would be a good answer to diversification is that network vendors have a choice of multiple suppliers – some proprietary and some open stacks. Let the best man win."
This sounds like eminently sensible advice from the former BT man. Whether the UK or other governments follow it is another matter. For some, the opportunity to create national champions that might eventually compete against Ericsson and Nokia in the global market may be too enticing. Economic realities mean few newcomers are likely to succeed.
At a country level, the frontrunners today are Japan and the US, two nations with a proud tech heritage. The former already had Fujitsu and NEC, equipment firms now building open RAN products. Home to the public cloud hyperscalers and semiconductor giants, the US can also boast the open RAN software developers whose names always surface in trials and tenders – Altiostar, Mavenir and Parallel Wireless.
Nothing prevents Ericsson, Nokia or other big players, such as South Korea's Samsung, from entering the fray. Indeed, all three of those vendors have signaled varying levels of interest in open RAN technology. They also have the resources to buy any open RAN startups that seem like a threat, much as Facebook swallowed Instagram and WhatsApp when they began to look dangerous. While that possibility could trigger defensive takeover moves by operators, they would be crossing the same threshold separating operator from vendor as Rakuten.
Amid neck-craning forecasts and reams of marketing hype, open RAN looks set to remain a discussion point for years. How open and diverse the market turns out to be is certainly up for debate.
- Vodafone UK to swap big part of Huawei for open RAN
- Open RAN might not be ready for America's big 5G push
- Open RAN and the Everest of inflated expectations
- Pret a open RAN
- The political hijacking of open RAN
— Iain Morris, International Editor, Light Reading