Floundering in open RAN, the US sniffs around Rakuten

US telcos have become heavily reliant on foreign vendors amid preachiness about open RAN.

US policymakers have a weird obsession with open radio access network (RAN) technology, a pick-and-mix approach to building mobile networks. Behind it seems to lie a protectionist ambition to have a Stars-and-Stripes kit vendor and not rely on foreign producers. The rationale is that open RAN – by ditching the proprietary interfaces that tie an operator to a single vendor for all RAN products – will spur competition. It has spawned a club called the Open RAN Policy Coalition, whose membership page reads mainly like a who’s who of US tech companies. But four years since that was founded, the US mobile market is less American than ever.

T-Mobile US, the fastest-growing operator, is majority owned by Deutsche Telekom, Germany’s telecom incumbent. More importantly, Sweden’s Ericsson now dominates the sale of RAN products to US operators. Once it has replaced Finland’s Nokia across a third of AT&T’s network, it will be the sole supplier of RAN and management software to that company. It is the biggest supplier to Verizon and provides about half the 5G kit for T-Mobile. This leaves Nokia and South Korea’s Samsung as the only other vendors with a sizeable US presence.

All this ignores Dish Network, a “greenfield” operator whose list of suppliers seems as long as War and Peace. But Dish looks in worse shape than a Napoleonic soldier on the Russian march. With about $20 billion in debt, it slid to a net loss of about $10 million for the first three months of 2024, down from a profit of $223 million a year earlier, despite making $3.6 billion in sales. It relies on the networks of AT&T and T-Mobile to provide services – effectively making it a user of Ericsson and Nokia products – and serves only 7.3 million mobile customers in a country of about 340 million people. Analysts expect it to go bankrupt.

Its open RAN, the construction of which explains Dish’s financial plight, features only one notable US RAN vendor – Mavenir. Funded by Siris Capital and Koch Industries, it provides the RAN software. But Dish’s radio units (RUs) are supplied mainly by Fujitsu of Japan and Samsung.

American impotence

Ironically, the US today has fewer homegrown RAN options than it did before the Open RAN Policy Coalition took shape – surely proof of that club’s impotence. Parallel Wireless, a Mavenir rival, reportedly laid off hundreds of employees in 2022 and has subsequently failed to land contracts. Altiostar, a third member of this software group, was acquired in 2021 by Japan’s Rakuten, which offers its products under the SymRAN brand. It is an understudy to Mavenir on the Dish project but so far appears to have been loitering off stage.

Representatives of Rakuten did, however, meet with various US officials from the Federal Communications Commission this week. And they included the top brass of Jessica Rosenworcel, the FCC’s chair, Rakuten Group CEO Hiroshi Mikitani (Mickey Mikitani, as he is known to westerners who presumably mangled Hiroshi too much for his liking) and Sharad Sriwastawa, the president of Rakuten Symphony, the part of the company that sells SymRAN.

What could these telecom luminaries possibly have had to discuss? According to a letter about the meeting penned by Sriwastawa, the focus was open RAN and Rakuten’s role in it. And that included a new scheme that involves making the SymRAN source code available to other companies.

This is not how RAN vendors typically operate. Ericsson, Nokia et al would normally develop the software, combine it with their other mobile network products and sell the whole integrated package to a telco. Open RAN, not to be confused with open source, simply unlocks interfaces such as the fronthaul link between baseband units (essentially servers hosting RAN software) and radios. In theory, that allows an operator to use software from one vendor with radios from another.

But Rakuten is now offering to license its RAN software to other vendors. The departure from convention was explained by Sriwastawa at this year’s Mobile World Congress. “We don’t have a huge sales team, and there are these RU manufacturers who are strong in their own territory, especially considering 5G enterprise use cases,” he told Light Reading. Systems integrators will also be able to purchase a license and bundle SymRAN with a third-party RU, he said.

Since then, Mikitani has suggested Rakuten might go even further. “We have RAN software which we have announced is going to be open and we’re having discussions with many people,” he said on an earnings call earlier this month (via a translator). “In the world of Linux, Red Hat has succeeded … In terms of mobile phone networks, we are going to pursue becoming the Red Hat of the mobile world.”

Red Hat, to the uninitiated, is a provider of cloud-computing platforms and an apostle of the open source creed, of which the Linux operating systems are among the best-known scriptures. Nothing is licensed. Instead, Red Hat earns money from the packaging and services it offers around the “free” code. By drawing on the comparison, Mikitani was implying Rakuten could do likewise – make its RAN software freely available and look to other sources of income.

Where did all the vendors go?

This could well appeal to US officials still attached to open RAN but fully conscious of US misfires in this market. Many operators still want to buy their RAN products from a single vendor, or at least have a single point of contact and entrust most business to that party. An American RU vendor or systems integrator could theoretically build a pre-integrated RAN product on top of SymRAN instead of attempting to develop its own software from scratch.

One question is how, exactly, Rakuten would make money. Red Hat effectively contributes to and makes withdrawals from open source libraries to build the virtualization and cloud platforms it sells to its customers. If Rakuten chose to open source its SymRAN code, and was no longer performing the vendor or systems integrator job for a telco, it would need an alternative revenue source. What that might be is not obvious.

As noted, there is also a dearth of American RU vendors and systems integrators commercially active in mainstream mobile networks. Mavenir is probably the most successful. Besides working for Dish, it has landed a deal to sell radios to some of Deutsche Telekom’s European operations as well as a systems integrator contract with Virgin Media O2 (VMO2) in the UK.

But Mavenir already has its own RAN software. What’s more, the VMO2 deal looks small and Deutsche Telekom has had little to say about its work with Mavenir. Inside and outside Germany, the telco remains heavily reliant on traditional kit vendors, including China’s Huawei, a company still publicly opposed to the open RAN concept.

To critics, Rakuten’s strategic pivot will smell of desperation. Symphony’s revenues fell 17% last year, to $393 million, a fraction of the $17.8 billion that Ericsson made at its networks unit. Outside Japan and the US, its only notable deal is with Germany’s 1&1, another greenfield that has struggled to activate sites.

The bigger problem for Rakuten and other open RAN hopefuls is the market downturn. Revenues generated by RAN product sales fell 11% last year, according to Omdia, a Light Reading sister company, and it projects another decline of between 7% and 9% this year.

A shrinking market always suits the largest players with the most resources unless a challenger has something genuinely disruptive – a digital camera when everyone else is still using Kodak film. Open RAN hardly qualifies. It is a tinkering with the method, not a technological breakthrough. And it is available from the incumbents. If it weren’t, AT&T would not be insisting its $14 billion deal with Ericsson is open RAN-compliant. For all the chatter about supplier diversity, and having domestic alternatives, CFOs prefer to sign contracts with trusted and profitable giants than untried and loss-making new entrants.

Coda

On that matter, defenders of the open RAN faith reject claims the Dish Network features only Mavenir as a US vendor. Strictly speaking, they are correct. AWS and Broadcom-owned VMware provide cloud and virtualization technologies. Dell and Intel stump up chips and servers. Others are in the mix. Yet to accept these as viable candidates is to maul the classic definition of open RAN. A network without an open fronthaul interface can just as easily be deployed in the cloud on purportedly general-purpose chips and standard servers.

Dish’s open RAN, at least, ticks the “multivendor” box by combining Mavenir’s software with radios from Fujitsu and Samsung. But the use of Dell, Intel, AWS, VMware and other IT and cloud suppliers does not make a network a multivendor “open RAN.” All those companies can figure in a deployment whether it uses open RAN specs or it doesn’t. A traditional Nokia RAN could be deemed multivendor because it features components from Broadcom, Intel and Marvell. But no one would use that label. And it certainly wouldn’t be an open RAN.