Investment in the mobile network usually follows the peaks and troughs of a radio wave, up for the initial rollout of a new generation, down when it’s mature. But 5G spending has the downward slope of a plane’s flight path on its final descent, and the job looks far from done. In the UK, market leader BT covers only three quarters of the population, and indoor signals are often weak. The squeeze has lasted more than a year, longer than expected by Ericsson and Nokia, the main vendors. Worryingly, there might not be a rebound.
Dell’Oro and Omdia (a Light Reading sister company), two analyst firms, have already forecast another sharp fall in telco spending on mobile network products this year after the big dipper of 2023. Now Analysys Mason has weighed in with a longer-term view on overall network spending. It is a gloomy read for anyone who sells network products to operators, warning that a “long decline” in capital expenditure (capex) has now started. “There will not be a cyclical recovery,” says one subhead.
Analysys Mason crunched a lot of numbers to arrive at this conclusion, processing historical data for about 50 of the largest operators in the world. Importantly, it also looked at the long-term guidance issued by those companies. Capex has peaked partly because telcos in many regions have completed or are near completing a once-in-a-lifetime upgrade to full-fiber networks. Clearly, that’s bad news for companies selling the actual fiber. Operators will continue to invest in the active electronics for these lines, but that represents a “tiny fraction” of the initial cost.
‘We’re not doing that again’
This is unlikely to hurt Ericsson, which has no fixed-line business, or Nokia, which deals in active electronics rather than passive equipment. The worry for the Nordic vendors is Analysys Mason’s take on 5G standalone and 6G capex. The former will be less capex-intensive than its non-standalone predecessor because it is mainly about the replacement of the software “core” rather than the installation of new radio access network (RAN) kit. But Analysys Mason also thinks “slack demand” will make justifying another big upgrade very hard for most telcos.
The rollout of 5G has not, of course, produced any new, monetizable services that telcos can offer their customers. Instead, it is merely succeeding 4G as a high-speed mobile data connection for smartphones. In competitive markets, telcos offer such generous bundles of gigabytes for set monthly fees that higher usage makes no difference to revenues. Meanwhile, competition is perennially forcing down prices.
As for 6G, outside parts of the vendor community, enthusiasm for a new standard in 2030 or thereabouts is low. Howard Watson, the chief security and networks officer of BT, looked horrified at last year’s Mobile World Congress when he was asked if 6G would trigger another big equipment upgrade. “We’re not doing that again,” was his quick response.
Since then, operators in the club of the Next Generation Mobile Networks Alliance have joined him in protest. “6G must not inherently trigger a hardware refresh of 5G RAN infrastructure,” they say in a report. Where possible, those operators want to be able to introduce 6G via “software-based feature upgrades of existing network elements.” Coming after this, the assessment of Analysys Mason is that 6G, like standalone 5G, “will not be capex-intensive” because the appetite for it just isn’t there. “There will be no cyclical uplift,” says the report.
Accordingly, the analyst company reckons capital intensity, which measures expenditure as a percentage of revenues, is due to sink. Across the operators it examined, capital intensity is today about 20%. By the end of the decade, the figure will have dropped to between 12% and 14%, it predicts.
Support for this can be found at a couple of telcos. BT, notably, has said it is beyond “peak capex” and expects to cut about £1 billion (US$1.3 billion) off annual investment, which came to £4.8 billion ($6.1 billion) last year, by the end of the decade. Spain’s Telefónica, having extensively built out fiber in its domestic market, expects group-wide capital intensity to fall from about 13% now to less than 12% over the next few years.
Data plateau
Much of this is at odds with the messaging put out by telcos about “exponential” growth in data traffic that forces them to continually invest bigger sums. Recently, it has been seized on by the bosses of Ericsson and Nokia. Have patience, they say to investors, because that rising tide of petabytes will eventually submerge networks that resist upgrades. Telco customers will have to spend.
Yet there are various problems with this analysis. For one thing, growth in data traffic is no longer exponential, whose strict definition means the rate of increase is going up from one year to the next. Data from various analyst companies, including Analysys Mason and Omdia, shows that in key markets growth rates are actually in decline. William Webb, an independent consultant and former executive at UK telecom regulator Ofcom, forecasts an S-curve flattening by 2027.
In a forthcoming book called “The end of telecoms history,” Webb returns to predictions he first made in 2016 to gauge their accuracy. Using recent historical data from Barclays, he was able to show a close alignment with the S-curve he drew about eight years ago. If this behavior continues, growth rates “will fall to near zero by around 2027, with significant variations by country,” says Webb in his book, giving a sneak preview to Light Reading.
Webb’s broad rationale is that there is an upper limit on daily gigabyte consumption, just as there is only so much the average person can eat or drink. All Webb had to do was assume there will be some future gorging by customers on high-quality video, the most calorific meal for any smartphone. “Once they are watching video for all their free moments while downloading updates and attachments there is little more that they could usefully download,” he writes.
What of future services people do not currently enjoy? Outside virtual reality – which, for safety reasons, will probably always happen in a fixed-line environment – no app seems likely to chew through gigabytes as hungrily as moving images do in high definition. Webb clearly doubts the sort of artificial intelligence (AI) services being advertised by Apple will have much impact whatsoever.
“There may be substantially more traffic between data centers as models are trained but this will flow across high-capacity fiber connections which can be expanded easily if needed,” he told Light Reading by email. “At present AI interactions are generally in the form of text, which amounts to miniscule amounts of traffic.”
“Indeed, if time is diverted from consuming video to AI interactions, then AI may reduce the amount of network traffic,” he continued. Even if AI is used in future to create images and videos, rather than words, it will probably make no difference given the amount of video already consumed, merely substituting for more traditional forms of content, said Webb.
For those confident that data traffic growth stimulates investment, the other problem is the lack of any correlation between volumes and costs. Advanced networks are designed to cope with usage up to a certain high threshold before an upgrade is needed. Headline expenses have not risen in lockstep with gigabytes.
Take Telefónica, for instance. The volume of traffic managed on its global networks rocketed from about 17,000 petabytes in 2015 to 146,000 last year. Yet over that same period, annual energy consumption – a proxy for operational costs – has fallen from about 6.6 million to 6 million megawatt hours. In 2015, Telefónica spent €37.6 billion ($40.3 billion) on supplies, personnel and other expenses. Last year’s operating costs were just €30.8 billion ($33 billion). Capital intensity has fallen from 17% to 14% over this period.
Irrelevant metrics
Despite all this, parts of the industry remain fixated on metrics such as “cost per bit.” A focus on driving that down is one of the broad strategic options for today’s telcos, according to Graham Baxter, a director at consulting firm Grey Bee who previously worked in senior technology roles for telcos such as Three and Orange. “Ultimately, that gives you the biggest profitability,” he said at an event in London hosted by Rakuten last week. “If you want to maximize your profits, you’ve got to lower your costs.”
But it’s an irrelevant metric, according to Dean Bubley, the founder and director of Disruptive Analysis, who says a wholesale network operator like CityFibre would not even know how many bits a retail customer such as Vodafone is pushing over its network. Cost per bit would be huge on a newly constructed network with hardly any traffic, falling dramatically as customers are added without making any real difference to overall expenses.
At some point, cost per bit will inevitably hit a floor, too. Margins will be squeezed if prices are chasing it down at a faster pace, and declines have been sharp in competitive places. In the UK, for instance, Vodafone’s average revenue per user for a contract customer plummeted from £26.70 ($33.80) in the final quarter of the 2015 fiscal year to £17.20 ($21.80) in the most recent one.
The floor might not be far below current cost levels, either. The world’s biggest telcos have already scrapped hundreds of thousands of jobs in recent years, outsourcing roles, abandoning asset ownership for tenancy, sharing equipment with one another. AI could always help with efficiency, but Webb urges caution.
“Some believe it could be used to learn how best to optimize each individual cell in a mobile network for greatest capacity,” he said. “However, there are numerous obstacles to be overcome before this can occur including finding ways to train the system and ensuring this does not result in network failure through unforeseen errors or unanticipated interactions with other intelligent optimizers within the network. And if mobile data growth is slowing then there is limited benefit in enhancing network efficiency.”
A much brighter future would await telcos if they were able to “create some level of innovation,” as Baxter puts it, and boost revenues. Ericsson and others are trying to spur sales growth by engaging with coders on 5G. Software developers, they hope, will pay for access to network features previously hidden away. The outreach could also engender new revenue-generating services. Many are skeptical.
Even if this happens, it does not guarantee business for equipment vendors. Analysys Mason sees many positives for telcos in a less capital-intensive future. “It could be a window of opportunity to shift their investment away from the old productive forces that deliver bandwidth at ever-lower cost, and instead harness new and higher-quality productive forces that exist outside those traditional perimeters,” it concludes its report. But its forecasts about capital intensity imply any revenue uplift for telcos will not enrich vendors.
At Rakuten’s London event, there was talk among analysts of operators sweating 5G assets for up to 15 or even 20 years, instead of upgrading networks every seven to ten years, as they would previously have done. It would make telcos look even more like other utilities with aging infrastructure. But no company will invest when the returns are poor. For suppliers now experiencing the steepest market decline in more than two decades, it could be a disaster.
This post was originally published on the 3rd party mentioned in the title ofthis site
