Network capex has bottomed out, but cost efficiency remains MNOs’ key driver

The past two years have been very difficult ones for major telecoms equipment vendors, with an almost global squeeze on operator capex.  In some regions, such as the US, capex is expected to rise in 2017, though forecasts by IHS Markit, and other analyst firms, suggest that global telecoms spending will not get back to 2014 levels (about $353m across fixed and mobile) until 2020, or even then. So while 5G, and fiber upgrades and expansions, will inject some new cash into the value chain, the telcos will remain heavily focused on cost efficiency amid falling ARPUs and uncertain 5G revenue models.

Technologies which use capacity more efficiently, and therefore reduce the need to upgrade and expand networks, will dominate procurement more than ever before. In many cases, even in ‘5G pioneer’ countries like the US, these will be 4G enhancement options, and deploying 5G commercially will not happen unless operators can justify it in terms of greater efficiency in spectrum, capacity and cost of ownership. While additional revenues, directly generated by 5G, will be important to the business case, many operators say they have not yet identified such use cases with any real confidence, so new service revenues are likely to be the second-stage return on 5G investment.

Efficiency must come first, and this will often revolve around fixed/mobile convergence. Verizon and AT&T are focused on fixed wireless for their initial 5G trials, which is partly a convenience – no need for a smartphone base to develop; a way to get round the regional fences around their fixed broadband markets. But this is also an early indication of how overall capacity efficiency will be greatly improved if traffic and usage can be seamlessly and flexibly routed around any kind of broadband connection, wireline or wireless, creating a single pool of capacity. Additionally, wireless can fill gaps in wireline coverage and reduce the total cost of providing universal gigabit access; while also supporting dense backhaul and fronthaul.

Of course, vendors will be well aware that greater efficiency, and pooling capacity across multiple networks, will keep capex levels relatively depressed. Convergence, 5G and fiber will drive more spending, but the potential will be lower than in the last major upgrade cycles, such as the first wave of 4G migration. Seizo Onoe, CTO of NTT Docomo, one of the 5G frontrunners, has emphasized that cost efficiency is central to the model, even ahead of new revenue potential.

He told conferences last fall that “better service does not always correlate with greater capital expenditures”. He pointed out that Docomo had spent less on 4G than on 3G, and expected to spend less again on 5G, partly because it would use the same infrastructure as 4G in many cases, and partly because of more efficient radio technologies and new network architectures such as virtualization. The operator’s 2015 capex bill of ¥600bn was a 15-year low, even though its data traffic was at the highest level ever and up 6,300% from 2000.

Other operators are less optimistic about 5G costs. In particular, companies like Deutsche Telekom have questioned the high expenditure which would be necessary to roll out 5G in millimeter wave spectrum, which would require large numbers of short-range base stations. And in the 4G migration, many operators, including those in China and much of Europe, found that upgrading to LTE did not enable them to increase revenues. If the costs are high, or there is no premium to command from 5G, most operators will not make the move, but will continue to squeeze more out of LTE, which has a significantly roadmap ahead of it.

The overriding drive for cost efficiency has helped create serious problems for large vendors in the past two years, highlighted by poor results at Ericsson, and some of the analyst reports which have come out in the early part of this year. For instance, IHS Markit calculates that revenues from mobile network infrastructure fell by 10% in revenue terms between 2015 and 2016, to $43bn, while the fall in macrocell equipment dropped even more steeply, by 14%. The decline was down to LTE slowdown and price reductions, with China leading the downward trend in spending. There was some offset from network software, mainly because of LTE-Advanced upgrades, but that segment was still up only 2% year-on-year.

The company predicts continuing LTE decline in 2017 and only a slow growth from 5G. It said: “All indicators point to a year of LTE decline in 2017 as a result of diminishing roll-outs worldwide. Moreover, we forecast the LTE market to decline at a compound annual growth rate of -12.4% from 2016, sinking to $12bn from its peak of $25.9bn in 2015. Early 5G roll-outs won’t be enough to push the overall market back into growth territory.”

This picture is not a simple one, however. Big deployments of LTE macrocells may have peaked, but operators are still spending. Now they are focused on software and antenna upgrades, small cell densification, and the start of virtualization, along with a few 5G radio trials. All these, as outlined above, are geared to increasingly capacity, but crucially, reducing the cost of that capacity. That may be bad news for the big telco suppliers, in terms of overall budgets, but it will be good news for companies working in more specialized areas which directly impact on efficiency – advanced antenna developers, for instance, as well as the companies which are focused on making the operator’s network both converged and flexible.

That means that one of the bright points in this dark picture will be software-defined and virtualized platforms which steer traffic and allocate capacity flexible between different network resources – wireless and wireline, transport and access. Transport companies like Ciena and Infinera have been making progress in this area and unlike the big OEMs, which also have offerings here – especially Huawei – their growth will not be offset by the difficulties in macro radio gear.

The pattern of capex spending is regionally varied. The great hopes which vendors placed in the BRIC (Brazil Russia India China) markets a few years ago have been dashed since 2015. Russian and Chinese operators will reduce capex again this year – MTS told its investors last week that it had cut capex by 13.1% last year, to RUB83.6bn ($1.45bn), and it was unlikely to rise this year. That market is facing economic pressure combined with over-capacity, while in China, the rapid 4G roll-out is now slowing amid price wars and falling profits for the big three carriers. Brazil is in the throes of financial upheaval, and though Indian operators are expanding their 4G networks at pace, they are very price-sensitive and are now going through a wave of consolidation.

With Europe, as Ericsson’s most recent results showed, also watching the costs amid economic uncertainty, vendor hopes will be heavily focused on the early 5G deployments in the US and parts of Asia, as well as newer 4G markets in Latin America and Africa.

In 2017, the big four US operators look the most robust in capex terms, and for the western vendors, there is the added attraction of not having to compete with Huawei for any gear which counts as critical infrastructure.

UBS analysts are predicting that US wireless carriers will spend a total of $31bn on their networks this year, “the first year of growth after three years of declines”, because of densification and new spectrum, as well as traffic growth resulting from the return to unlimited data plans. American Tower is predicting data traffic growth of 30% to 40%.

According to Wells Fargo Securities, US operators will increase their network spending this year, compared to 2016, driven by expansion into new spectrum bands – such as AT&T’s AWS-3, and even the first 600 MHz projects – and densification. The analysts predict that Verizon will increase wireless capex from $11.5bn last year to $11.9bn in 2017 as it densifies its network with small cells; while T-Mobile will continue to roll out LTE in its 700 MHz airwaves to cover up to as 270m POPs. AT&T has said it will reduce total capex from $22.9bn in 2016 to $22bn this year, but analysts expect the wireless portion to increase.

“We anticipate that wireless spending overall will accelerate in 2017, with the largest increases coming from AT&T and Sprint,” Jennifer Fritzsche of Wells Fargo wrote in a client note. “With a better wireless spending environment in 2017 vs. 2016, one obvious beneficiary should be the tower segment. Even as new tower builds are given to new/smaller developers, it is hard for us to foresee how the 96,000 towers, which the three public tower companies own, are not touched with cell site amendments.”

Last week, T-Mobile’s CFO Braxton Carter added weight to Wells Fargo’s assumptions, telling TMO investors: “There’s no catalyst for a decrease” in network capex. However, he also does not see “any catalyst for any significant step-function increase in our capex profile”. The first priority is LTE expansion in 700 MHz, and then the operator will look to other ways to increase capacity and efficiency. “Once we finish that project, there’s a pool of dollars that can roll to do other things,” he said. “There’s no reason that we can’t continue growing the way we’ve been growing.”

American Tower increased its 2017 outlook by $100m thanks to an amended master lease agreement with one tenant – believed by most to be AT&T – which it disclosed in a regulatory filing. It improved its property revenue outlook to a range from $6.31bn to $6.49bn.

“We suspect this agreement formalizes AT&T’s intentions to deploy AWS-3 and WCS spectrum later this year,” New Street Research analysts wrote in a note to investors. “This is a positive signal that US spending on towers is set to accelerate.” AT&T has about 40 MHz of unused AWS-3 and WCS spectrum. It is also likely to win the FirstNet public safety network contract, which would also boost its tower partners, since it would build a $6.5bn network in 20 MHz of 700 MHz spectrum, for use by federal, state and local authorities. AT&T will have the right to sell excess capacity itself.

Of the big four, Sprint has cut capex the most in recent years, and though it has argued that its plentiful supply of 2.5 GHz spectrum, and its advances in network optimization and MIMO, give its networks huge efficiency advantages, it has still suffered from delays and under-delivery. For many observers, these are signs that it has cut its budgets too aggressively. This year, it will be forced to boost its spending, although its fierce rivalry with T-Mobile, and the price wars this drives, may lead Sprint’s owner, Softbank, to reopen the idea of a Sprint/TMO merger, which would transform the cost efficiencies of the third and fourth MNOs.

While they remain separate players, Sprint is likely to pursue densification to support cost-effective capacity increase in its 2.5 GHz airwaves, and Wells Fargo is optimistic that some of the logistical barriers to large-scale small cell roll-out, such as municipal regulatory red tape, will be lowered during the course of this year, under pressure from the FCC and from the cities themselves, as they move towards being ‘smart’.

Sprint’s CTO John Saw told a recent conference that its investments in 8×8 MIMO have helped keep overall capex low, while other updates, like three-carrier aggregation, have been in software. However, he acknowledged that software cost savings will disappear and spending will increase accordingly when massive MIMO kicks in, adding that 64×64 MIMO is “not far away”.

Other operators and analysts echo the view that the limits of efficiencies from spectrum and software are being reached in developed markets, and operators will have to enter another cycle of more expensive hardware upgrades, in antennas, radios and – for virtualization – servers.

In a research note, MoffettNathanson analysts said spectral efficiency has improved significantly in recent years by harnessing carrier aggregation, MIMO and QAM, but operators may now be “approaching the theoretical limit of spectral efficiency”. This will make reuse of the spectrum they already have, through densification, increasingly important for adding incremental capacity, combined with continued advances in MIMO. “Network densification sets us up nicely for 5G, a next generation standard that will not only use the traditional frequency bands but also aggregate them with hundreds of MHz of millimeter wave spectrum,” they wrote.

Wells Fargo sees Sprint leading the shift back to advanced hardware and to spectrum reuse. “At Sprint, the carrier has delayed some capex in fiscal year 2016 due to the procurement of permits for its densification plans, so we expect fiscal year 2017 to increase from the $3bn or so it will spend in fiscal year 2016,” wrote Fritzche. “Given the fact Sprint’s liquidity position has significantly improved, we would not be surprised to see this ramp to $5bn-plus. The key question for us is, if Sprint can achieve such network improvement as it did in 2016 with low spending, what kind of advantage will they get when the real spending starts?”

In particular, Fritzsche is more bullish than most analysts about Sprint’s chances of catching up, because of its 2.5 GHz spectrum, which give it three advantages in the capacity/cost efficiency game – it will support migration to 5G, where 2.5 GHz could be a ‘low band’; it supports three-carrier aggregation in a simplified form, within a single band; and it makes it easier to deploy Massive MIMO because higher frequencies can work with smaller form factors.
Where Fritzche is more in line with the more Sprint-skeptical thinking of her peers is in recognizing the synergies which would result from a merger with TMO – “significant (massive?!)”, she wrote. However, if Sprint does stay independent, she thinks it “has a long runway even if on a go-alone strategy. Because of its spectrum advantage and technology advancements, when we think about strategic relationships across the wireless industry, we believe even in the absence of a merger with another player – S has more than enough in terms of spectrum and network capabilities to stand on its own two feet.”