For all the talk about open RAN—and there is a lot of talk—what is most telling to me is how little open RAN there is.
We have all heard about Rakuten’s network in Japan (which may not be so open after all) and Dish’s network in the US. Vodafone and Telefonica and other EU operators have signed their very public Memorandum of Understanding. Even the FCC is on the bandwagon.
My calculation, based on market forecasts I have worked with, is that the total number of open RAN radios shipped from 2021 through 2025 will not likely exceed 5% of the probably 50 million 5G radios that will ship world-wide in that period. Both figures include small cells, by the way.
In other words, looking five years out, open RAN really doesn’t matter.
Which is not to say it doesn’t matter to firms like Mavenir, and Fujitsu, and NEC. If Mavenir can get on track to work its products into even a few of the world’s mobile networks, their investors will finally see the payday they’ve been waiting for. It just won’t mean much in the grand scheme of things 5G.
So I think this question a fair one: What is the value of Open RAN? In particular, as defined by the ORAN Alliance.
The arguments in favor are usually, 1) more competition will lower costs, and 2) operators want a broader supplier ecosystem.
In my professional experience, (1) is bilge. The Big 3 (Nokia, Ericsson, and Huawei) have spent years ruthlessly cutting costs, painfully so for their suppliers. Being a supplier to Ericsson may not be quite as difficult as supplying Apple, but it isn’t much easier. Why? Because the Big 3’s customers are few, large, control huge budgets, and are extremely savvy. The Big 3 business model is not like Apple pitching eye candy to persuade consumers to cough up a mortgage payment for a phone. You get to sign a multi-billion dollar supply contract with Verizon only after a couple of years of trials and brutal negotiations. And with all the R&D spending needed just to stay in the game, Ericsson’s bottom line is hardly a juicy one. (Nokia’s is still underwater, I think.)
(If Fujitsu is selling radios to Dish at a much cheaper price than Nokia, then these are budget radios, the TracFones of the network. And for Dish this makes sense. Deploy as much coverage, as fast as possible, with the least capex, to keep the spectrum assets alive and their value multiplied by rights-of-way, back-haul, and cloud contracts. If the radios have to be pulled out in five years (likely), or the entire business sold off (possible), either way Charlie Ergen and his team can cash out handsomely. They will have done their jobs.)
The real answer to (1) is that the market for RAN solutions is already extremely competitive. People who don’t think so seriously underestimate it.
What about (2)? Well, history says it’s largely smoke. Motorola, Nortel, and Alcatel-Lucent folded years ago under ruthless competition and sold out. The operators, at least the major ones, want solid, reliable suppliers who deliver top value for the capex buck. The fact that Huawei has been taken off the table in the West may have strained the situation, but the solution for T-Mobile is not to start buying equipment from a motley collection of white-box suppliers.
Still, if there is one value proposition that open RAN might bring to the party, it could be to provide a lever the operators can use to break the vendor lock.
If you’re in the business, you’re very familiar with the vendor lock. Ericsson baseband gear only works with Ericsson radios, and Nokia with Nokia. As an operator, you build your access network in single-vendor islands or layers. There is no mixing and matching.
In reality, there is no mixing and matching in open RAN, either. Not now, and not in the immediate future. The main reason for this is that the ORAN specifications leave too much undefined, and each vendor must fill in. I would add, though, that the Alliance’s architecture was a necessary compromise, a collection of tradeoffs to get something off the ground. Where Ericsson can move functionality around between baseband and radio in order to optimize performance, the ORAN model is an architectural lock, and one that must be respected if there is to be mix-and-match in the future.
But a major operator might break that architectural lock. The ORAN front-haul spec could be bent a little to work with another functional partitioning, say to get more processing into a massive MIMO radio. And a big operator like Verizon is in a position to push vendors into a form of technical cooperation. (Rakuten did it with Nokia.) The ORAN Alliance moves at a glacial speed and is notoriously resistant to revisiting their model, but if presented with a field-tested modification to the published architecture, I think they would very likely be persuaded to adopt it without too much fuss. After all, Verizon is a founding member.
Is there truly value in breaking the vendor lock, aside from the psychological satisfaction? Very likely. There is a lot of mathematical, algorithmic, and multi-physics secret sauce in RAN solutions. If you totally commodify it, you get WiFi. And 5G is not Wifi.
While you can’t expect Vendor X to hand over its scheduler algorithms to Vendor Y, maybe you can get Vendor X to interface its baseband with Vendor Y’s massive MIMO. And if you have to bend the front-haul spec in the process, in the end everyone can still win.
Of course, you can argue that Verizon would not want to see its field-tested architecture shared with ATT. But it may be a necessary price, at least after a period of exclusivity, to make it worthwhile for their major suppliers to invest.
Looked at from a different perspective, open RAN could be a way for the major OEM R&D budgets to gradually become more specialized and more focused, rather than a way for any ol’ OEM to play in the RAN.
But, I suppose, also a way to sell budget-class gear.
If you have thoughts on this, please share.