A major telco gripe about open RAN is that its chip technologies are a poor match for the customized silicon that goes into traditional basestation equipment. Another is that Europe lacks much local chip-designing talent. Vodafone, one of the region's biggest operators, is now trying to fix both problems by opening its own silicon lab in Spain.

A huge talking point in the industry, open RAN – as a concept – is supposedly about unlocking the interfaces between different parts of a radio access network (or RAN). Achieve that, and operators might not have to buy all those parts from the same supplier's system, as they mainly do today. A related development is "virtualization," allowing RAN software to run on general-purpose server equipment rather than dedicated hardware. Whatever the attractions, an x86 server processor made by Intel is a weak substitute for an application-specific integrated circuit from Ericsson, Huawei or Nokia, say the experts.

This performance lag is one reason operators are wary of deploying open RAN in busy urban areas. The answer appears to be more semiconductor investment. Besides x86 processors, operators want access to more advanced merchant silicon they can use in open RAN deployments. Marvell, Nvidia and Qualcomm are just three companies hoping to satisfy this need with forthcoming products.

Figure 1:

An 8-inch silicon wafer inside one of TSMC's fabs in Taiwan
(Source: TSMC)

Vodafone's decision to open an R&D center in Malaga, announced months after those chip designers first mentioned their open RAN plans, is curious. The operator's official release is understated. Just 50 open RAN experts at Vodafone will join another 650 software engineers, architects and technicians at the facility, it says. They will, importantly, be working alongside about 20 chip vendors, Vodafone goes on to explain.

But a Reuters report on the move suggests Vodafone is turning silicon designer. That seems to overstate Vodafone's role. It has no real experience at a company level of designing chips and probably lacks the resources to make a serious commitment to silicon research and development (R&D). Qualcomm alone had $7.1 billion in research-and-development expenses last year.

Moreover, the companies branching into chip design are typically US tech powerhouses such as Apple and Amazon, with enormous R&D budgets. Apple profits by weaning itself off chip suppliers like Intel and producing more advanced gadgets for its customers. Amazon, similarly, benefits through improvements to its cloud-computing infrastructure.

Hard sell

A silicon foray by Vodafone would be a much harder sell to shareholders. The goal would presumably be to lower the cost and boost the performance of open RAN. But Omdia, a sister company to Light Reading, reckons overall sales of RAN equipment peaked globally in 2021 and will decline for the next few years. Regardless, operators make their money from services, not equipment sales, and there is no logical reason for the rollout of open RAN to spur growth in service revenues. Meanwhile, the argument that open RAN will produce a more diverse supplier market for operators is hotly disputed and unlikely to excite any telco investor.

Far likelier is that Vodafone aims to ensure chip vendors are meeting its requirements. There will not be a Vodafone chip but a Qualcomm one built to Vodafone specifications. The operator had already announced a tie-up with Qualcomm last year to produce "reference designs" for open RAN products. The Malaga initiative seems like a follow-up.

The need for alternatives to Intel seems paramount. Santiago Tenorio, Vodafone's director of network architecture, is cited in the Reuters article saying that Intel is up to three years ahead of rivals in open RAN development. But a market dominated by Intel hardly sounds preferable to the RAN oligopoly of Ericsson, Huawei and Nokia. And open RAN vendors will not look competitive equipped with only x86, said Scott Petty, Vodafone's chief digital officer, in late 2021. "Now they need to deliver, but it will require some dedicated silicon. It won't be Intel chips."

While Intel is named by Reuters as one of the 20 vendors active at the Vodafone center, others include Broadcom, Qualcomm and Lime Microsystems, a UK-based company that counts Tenorio as its chairman. Another is Arm, the UK-headquartered firm that sells blueprints to a long list of chip developers, including Amazon, Broadcom and Qualcomm. Vodafone's center will design silicon based on not only x86 and Arm but also RISC-V, an open-source alternative, Tenorio told Reuters.

European lag

Is any of this strictly necessary? Arm's customers, as noted, already have products in development. Arm, meanwhile, is opening a 5G lab in New Jersey with an open RAN focus. It features numerous silicon partners, just like Vodafone's, according to an announcement made in October. And Vodafone is involved there.

The primary concern for the operator seems to be that chip development is mainly done outside Europe. At a broader level, European authorities seem equally worried. One goal announced last year is to double Europe's share of global semiconductor production to about 20% by 2030, the Reuters article points out.

But the Vodafone R&D center will make no difference here, simply because it is not a chipmaking facility. Vodafone will reportedly invest as little €225 million ($269 million) in the Malaga center over a five-year period. To put that in perspective, Intel plans to spend between $25 billion and $28 billion on capital expenditure this year, up from $14.3 billion in 2020, while Taiwanese foundry TSMC is budgeting for $30 billion.

Nor is it clear how a Vodafone center full of American chip companies will give Europe more control. Although half the 20 vendors are said to be European, the muscle will mainly be American. If it were otherwise, Vodafone and its peers would not have repeatedly complained about Europe's shortcomings.

Persuading the Americans to invest in Europe is surely better than having no significant local ecosystem whatsoever. And yet this was already happening. "Qualcomm has put R&D in France," said Arnaud Vamparys, Orange's senior vice president of radio networks, when the French operator unveiled an open RAN test center in Paris last November. "We are in contact with the five majors that are investing right now," he added, referring to various US chip developers involved in open RAN.

Efforts by Vodafone, Orange and other service providers in Europe cannot simply be dismissed. Partnerships and knowledge exchange are required if an open RAN community is to succeed. Without crucial input from operators (and prospective customers), chip designers might overlook some needs. Just don't expect a thriving European ecosystem for open RAN chips to emerge from Vodafone's latest initiative.

Related posts:

— Iain Morris, International Editor, Light Reading