Nokia’s announcement of a $3.5bn 5G network deal with T-Mobile USA will be music even to the ears of its competitors, because it reinforces the notion that at-scale 5G RAN spending will start at an earlier stage than was earlier assumed.
It is certainly true that some operators are bringing forward deployments of initial 5G New Radio Non-Standalone (NR NSA) networks, under shareholder and regulatory pressure to act quickly on 5G, and to seize any first mover advantage that may be possible. However, Rethink Research’s most recent operator survey, which questioned senior executives from over 70 Top 200 MNOs, indicated that, as usual at the start of a new generation, the headlines are dominated by a few fast movers, and actually, the majority of operators which deploy in 2019 and 2020 will do so at small scale.
So this will not be a capex windfall for the OEMs. A new forecast from ABI Research estimates that the global RAN equipment market will grow at 5% a year between 2018 and 2023 to reach over $26bn at the end of that period. That is above the forecasts by the top three OEMs, which talk about a flat market now and growth rates of 2% to 3% in 2019-2020.
But the initial optimism that the ABI report may have sparked in the OEMs will be diluted when the details become clear – almost all the growth is in indoor equipment, particularly small cells, where the OEMs’ success is not guaranteed. Meanwhile, while it remains very hard for any new entrant to penetrate the highly complex macro network platform – in which Ericsson, Huawei and Nokia have invested billions of dollars in R&D – prices here will be squeezed as operators shift to open virtualized architectures and densification.
There are various factors which mean the big three may not dominate 5G in the same way they have the previous cellular platforms, and will not get the same spike in network revenue. The 5G deployments will typically be more stretched-out than those for 4G, and many operators will be modernizing their networks (using existing sites, activating ‘5G-ready’ base stations) rather than installing brand new RANs. The modernization trend hit Ericsson’s results in the early years of 4G, and the same trend will persist with 5G.
And even when mass deployment of new RANs starts, operators are playing down the idea of a capex spike. China Mobile, which promises one of the world’s largest 5G roll-outs, and is going straight to Standalone, has said it expects its deployment to be spread over many years, with most installations only being made when there is a clear demand. NTT Docomo’s CTO Seizo Onoe has been vocal in arguing that the Japanese MNO will spend significantly less capex on 5G than it did on 4G, thanks to the price wars in base stations; the virtualized, software-driven architectures in which it has been an early mover; and more open platforms.
At the 5G World Summit in London earlier this year, Swisscom and Three UK also played down hopes of a 5G-related boost in capex, largely because they do not believe 5G will create a significant boost in spending by their own mobile consumers. Three UK’s CTO, Bryn Jones, insisted – in a clear message to suppliers – that he saw no reason for a major capex increase, saying: “Networks are like painting the Forth Bridge. You roll out 2G and as soon as that is rolled out you start rolling out 3G. The capex is really substitutional and you manage within the envelopes.”
His counterpart at Swisscom, Heinz Herren, also said 5G could be managed within today’s “capex envelope”, adding: “With 5G there might be some incremental capex but if we are clever and add radios that can deal with 4G and 5G we will not see a big uplift in capex.”
He added that the business case still had to be made for at-scale roll-out, saying: “The Swiss market is flat and so it is really difficult to add 5G and sell it and personally I think we are having too many discussions around the business case. I think if your main business case is connectivity there is no question about doing 5G, but I don’t think you will see additional ARPU.”
They were supported by Constantine Polychronopoulos, CTO at virtualization company VMware, who is just the sort of supplier to strike fear into the OEMs’ hearts. Because virtualization and software-defined networking (SDN) have started life in the enterprise data center, entrenched suppliers there, such as HPE, VMware and Red Hat, have high hopes of crossing over into the previously closed mobile operator market and seizing some of the most valuable pieces of the 5G network value chain. And to make their case stronger, they are well used to monetizing open source platforms, which are only just starting to impact the RAN, and which will drive Ericsson and the others to make the kind of painful adjustment in pricing and profits that enterprise vendors did a couple of decades ago.
“We are moving toward software-defined radios and eventually upgrading a radio will just be a software exercise and a lot less costly,” Polychronopoulos said. “Virtualizing the network will address capex as well.”
The main areas of increased investment will be in extended fiber backhaul and fronthaul (though for many that will be leased, and so be an opex cost); small cells, to support indoor coverage and densification, and help MNOs harness high frequency spectrum; and more sophisticated antennas, including Massive MIMO. Those antennas should be a boost for the big OEMs and remain a sector which is closed to most new entrants. However, open interfaces will potentially allow antenna specialists like Kathrein and CommScope to deal directly with operators as well as via the OEMs; and there is even the prospect of open antenna platforms, such as Facebook has been pushing within its Telecom Infra Project (TIP) with the Aries open hardware developments.
“When you look at Massive MIMO, those antennas are heavier and bigger and there is more spending in that area than we had in moving from 3G to 4G,” Jones said. “There is also a much higher requirement for fiber.”
The bigger deals are likely to come when operators start to virtualize their RANs, which for many will go hand-in-hand with 5G Standalone, which then requires investment in the 5G core. For most, these major architecture changes are a few years away, and they will not boost the traditional RAN equipment market – in fact, they are designed to reduce capex and opex by virtualizing as many digital baseband functions as possible on commodity servers and white box switches, leaving stripped-back radio antenna units at the cell site. Worse still for the major vendors, operators are determined to see a truly open interface between those remote radio units (RRUs) and the virtualized baseband – not the semi-proprietary CPRI which currently supports fronthaul between RRUs and centralized baseband units (BBUs).
This is the goal of several open projects, which are looking to create standardized fronthaul specifications. The IEEE’s Ethernet 1914 TSN (Time Sensitive Networking) standard will underpin many of these.
One of the most influential efforts is that of the xRAN Forum, soon to be merged with the C-RAN Alliance to form the ORAN (Open RAN) Alliance. It announced its first set of specs, the xRAN Fronthaul Specification Version 1.0, in April, and has now followed up with Release 2.0 to fill in some of the gaps in that initial interface, particularly a weakly defined management plane.
So, the Forum has unveiled the xRAN Fronthaul Control, User and Synchronization (CUS) Plane Specification Version 2.0 and the xRAN Fronthaul Management Plane (MP) Specification Version 1.0.
The second version of the CUS plane spec includes support for two radio categories, A and B, to enable both simple and more complex functionality with broadly the same interface specification. It also includes support for additional LTE features like LAA (Licensed Assisted Access) and NB-IoT.
The first version of the management plane spec provides an open multivendor M-plane model for radios based on standardized protocols like NETCONF/YANG. It includes a comprehensive YANG model developed for LTE and 5G radio.
The group said the new definitions should allow a wide range of vendors to develop innovative, best-of-breed RRUs and BBUs to support different deployment scenarios, while allowing easy integration with virtualized infrastructure and management systems using standardized data models. That would get away from the vendor lock-in enabled by non-uniform specs like CPRI, while also encouraging a wider variety of RRH types to support the growing number of use cases and deployment topologies envisaged for 5G.
Some vendors believe there could be commercial xRAN products as soon as the first quarter of 2018, though Sachin Katti, a Stanford University professor who is director of the xRAN Forum, did not speculate on timescales, though he said the merger with ORAN should be finalized this month.
ABI predicts RAN growth will be almost entirely indoors:
According to ABI Research’s latest forecasts, the base station equipment market will be worth over $26bn in 2023, representing a CAGR of 5% between 2018 and 2023.
Within that figure, the main growth will be in indoor equipment, including small cells and DAS (distributed antenna system). Today, indoor equipment accounts for 27% of total RAN sales, but this segment will grow at a CAGR of 15.5% over the next five years to reach 42% of the total in 2023.
Nick Marshall, ABI’s research director, said: “Today the RAN equipment market is undergoing multiple technology transitions as network operators move to densify macro networks with small cells, tackle in-building wireless and evolve to new technologies such as 5G, Licensed Assisted Access, unlicensed and shared spectrum technologies such as OnGo in the United States, and MulteFire. These transitions are occurring against a backdrop of continuous technology evolution as networks upgrade to include MIMO, Massive MIMO, 256 QAM and carrier aggregation.”
Meanwhile, the US operators’ association, CTIA, is also feeling confident about small cells, as barriers to deployment, such as over-complex processes for site and equipment approvals, start to be addressed. “The number of small cells deployed is predicted to rapidly increase over the next few years from about 13,000 small cells in 2017 to 86,000 this year—a 550% increase—and over 800,000 by 2026,” CTIA said in a survey of the US market.
Analysts at Wall Street firm Macquarie Research predicte that between 5% and 10% of US MNOs’ operating expenses this year will go on small cells. T-Mobile USA plans to deploy 25,000 small cells in 2018 and early 2019, “on top of the approximately 18,000 small cells and DAS nodes already rolled out as of the end of 2017”.
Verizon said recently that 62% of its wireless deployments in 2017 were small cells, “a figure that will only grow larger as we deploy 5G in 2018 and beyond … Small cells are needed to meet exploding consumer demand for data, drive innovation, create new jobs, and fuel new services and capabilities such as smart communities, connected cars, smart farming, and the IoT.”
Crown Castle’s small cell business on track to be one-third of total revenues:
Towerco Crown Castle made an early bet on US densification and now has 60,000 small cell in its portfolio alongside its 40,000 macro towers. “Since we made our initial investment in small cells, we have seen the market rapidly evolve from a small opportunity and only a few locations to where we are today with all four of the major wireless customers deploying small cells at scale across all of the top markets,” CEO Jay Brown said earlier this month during the company’s quarterly earnings call.
He added: “We believe we are at the very beginning of what will ultimately be an opportunity that rivals or exceeds what we have seen play out with towers over the last two decades where demand has far surpassed what even we could imagine at the time we made our initial investments.”
Crown Castle executives added that, a few years ago, they were installing between 5,000 and 7,000 small cells a year, but are now at a run rate of between 10,000 and 15,000 small cells a year. The towerco signed the same number of new small cell bookings in the first quarter of 2018 that it did during the whole of 2016.
Even with just a single tenant, Crown Castle is recording yields of 6% to 7% on a small cell site and is starting to add a second tenant to some deployments, which boosts the yield to the “mid to high teens”. Brown said: “There’s a real benefit to the customer of sharing that infrastructure. I don’t see any scenario where the cost to construct comes down dramatically from where it is today such that the shared model is not the lowest cost alternative. So, our job day-in and day-out here at Crown Castle is to provide infrastructure at a much lower cost to our customers than what they could do on their own.”
Brown said that Crown Castle’s small cell business is on pace to reach $55m in revenues this year, up from around $40m last year, making it about half the size of the macro tower division.