Why 5G in the UK is so bad

A T-Mobile commercial currently showing in Slovakia features several households where small children look crestfallen when bad Wi-Fi seemingly freezes an online game. "Telekom," comes the desultory explanation from parents or exasperated outburst from siblings.

This is not, however, some weird admission of responsibility by T-Mobile. A feature of the service provided by the incumbent, which bought a majority stake in Slovak Telekom years ago, allows parents to choke off the Wi-Fi on deciding the offspring has gorged enough.

Tech-savvy kids, though, could possibly decamp to 5G. In the small town of Trstena, nestled in the foothills of northern Slovakia's Tatras mountain range, an iPhone user can enjoy a full four-bar 5G connection in the indoor restaurant of the Hotel Rohac, whose apple strudel and low-cost pilsner also receive full marks.

This would not be the typical experience for many visitors from the UK, where decent indoor 5G is rarer than a ghostly encounter. After one press conference at BT's headquarters in recent memory, reporters thumb-wrestled with Wi-Fi logins and passwords in the cellular drought of a local pub.

Via its EE brand, BT provides the UK's best 5G service, according to various measures. But the country seems to fare badly in a European region that itself trails other parts of the world. In a ranking of 25 European countries earlier this year, Opensignal, an independent monitor, put it fourth from bottom for 5G download speeds, with 118.1 Mbit/s. That was 25.5 Mbit/s less than a user would get in Slovakia, according to the ranking. In Denmark, the leader, 5G download speeds exceed 300 Mbit/s, said Opensignal.

Related:Vodafone UK Not Radio Gaga About Nokia

Much worse, perhaps, is the dire coverage. In October 2014, just two years after launching 4G, EE was boasting population coverage for that technology of more than 75%. But it has taken the operator more than four years to reach 74% for "outdoor" 5G coverage since its launch of the network in late 2019, according to data published by Ofcom, the UK regulator, for January this year. And other networks are much patchier, with Three on 67%, Vodafone on 57% and Virgin Media O2 on just 51%.

The independent monitors score networks even lower. Rootmetrics, an Opensignal competitor, recently said EE and Three "topped 60% 5G availability" for the first half of 2024, putting Virgin Media O2 (VMO2) on 55.9% and Vodafone on 47.2%. These figures and Ofcom's, though, are considerably higher than numbers from Opensignal published in late 2023. 5G availability was just 10.6% for EE, falling to 10.3% for Three, 10.1% for VMO2 and 10% for Vodafone.

The disparity is explained largely by methodology, it seems. Rootmetrics appears to ignore background smartphone tasks that run while the user is sipping a pint or bashing out a 5G story. Those tasks are often "managed over" to the 4G connection, it says. It also does more testing outdoors, where coverage is unimpeded by walls, furniture and stuff.

Opensignal, by contrast, says most of its data is gathered indoors because "that's where users spend most of their time." It also measures "the proportion of time people have a network connection." The prevalence of Wi-Fi connectivity may be responsible for the low numbers.

These are somewhat lower than Opensignal's 5G availability metrics for other European countries. O2, Slovakia's best network on this measure, scored 12.2% in a March assessment. T-Mobile in Germany, its domestic market, managed 15.9%. Networks in high-scoring France ranged from 16.2% to 18.2%. Denmark's TDC was on 17.1% in February.

The real gulf, though, is between Europe and other parts of the world rather than the UK and other European countries. Fast 5G networks in South Korea scored an average of 39.5% for availability in December, according to Opensignal. And while the US generally does badly on this measure, T-Mobile was at a stratospheric 67.9% in July.

Sob stories and Chinese trouble

Nevertheless, all this would seem to make the UK one of the worst-performing countries in a badly performing region. And a combination of factors is probably to blame. The disputed one it appears to share with various other European countries is the competitive intensity in what is a relatively small market.

In each of China and the US, just three big networks cater to a population in the hundreds of millions. In the UK, by contrast, four serve fewer than 70 million people. For a long time, parts of the European industry have argued this is one too many for any given country.

It is the justification for the perpetual sob story told by Vodafone and Three, which have insisted their weighted average cost of capital exceeds their return on capital employed. In those circumstances, no sensible investor will keep pumping money into 5G rollout.

A proposed merger between the two operators would supposedly fix this problem, ensuring the UK's pool of subscribers (and the money they spend) is divided between fewer networks. If competition authorities allow it, Vodafone and Three have promised to invest £11 billion (US$14.5 billion) in 5G rollout over the subsequent decade.

But not everyone buys it. Organizations that suddenly face less competition are not typically inclined to bolster their appeal. The UK had four networks during the rollout of 4G, which, as noted, was deployed much faster. An investment of £11 billion over a ten-year period would equate to just 11.7% of combined revenues when annualized, far less than the capital intensity of 25% that BT has recorded in recent years.

Most of BT's investment, however, appears to have been diverted into spending on the rollout of full-fiber networks. Until quite recently, the UK was stuck on the full-fiber start line, covered in the dust kicked up by European leaders such as Spain and Portugal.

Amid government concern, this all changed when capital investments began flooding into alternative networks, determined to build fiber in BT's backyard and poach its customers. New regulations gave them easier access to BT's ducts and poles. BT's response has been the extension of fiber to about 14 million premises, from just 1.2 million in March 2019. But the effort has conceivably pulled resources away from other activities, including 5G.

Operators in the UK, unlike in most other countries, have also had to contend with a government-mandated swap-out of Huawei, a Chinese vendor once perceived to be at the forefront of 5G technology. For reasons both technical and commercial, this swap-out, ordered after the US imposed sanctions on Huawei, has meant replacing 4G equipment, too. It is still not finished – with operators given until the end of 2027 to phase out Huawei's 5G basestation kit – and has inevitably gobbled up resources.

Yet its impact can be overstated. 5G had not been widely deployed in 2020, when UK authorities first moved against Huawei, and existing 4G networks were anyway due for replacement, according to some analysts. The £500 million ($660 million) that BT estimated in Huawei swap-out costs is a tiny percentage of capital expenditure over multiple years.

Even so, the replacement job has certainly not helped 5G rollout in the UK. Concerned about overreliance on a kit duopoly of Ericsson and Nokia, and apparently unhappy with Nokia's technology, Vodafone instead aims to replace Huawei at 2,500 sites – about 14% of its total – with an experimental system dubbed open RAN, provided largely by Samsung. But it did not begin its commercial rollout until a year ago, nearly three years after first revealing its plans. Much of the intervening period has been spent on trials.

Unspectacular spectrum

Spectrum and site density are also likely culprits for the 5G lag in the UK and various other European countries. When EE first began deploying 4G, it did so with 1800MHz spectrum, a range that sat comfortably between the signal propagation characteristics of lower bands and the fast speeds conferred by higher frequencies and their fatter channels. But the bands used with 5G are generally either too low for speed or too high for coverage, falling below 1GHz or above 3GHz.

The equipment industry has concentrated heavily on 3.5GHz and its immediate environs – referred to as the C-band in the US – describing this as a "midrange" sweet spot for 5G. It is anything but, according to some analysts. "Put simply, C-band isn't very good spectrum," said Craig Moffett, an analyst with MoffettNathanson Research, in a note published about Verizon's recent second-quarter results. Moffett clearly blames reliance on C-band spectrum for the appalling 5G availability metrics that Opensignal attaches to both Verizon (7.7%) and AT&T (11.8%).

T-Mobile appears to have overcome this problem by using a mixture of 600MHz and 2.5GHz frequencies to support 5G, after picking up the higher spectrum with its takeover of Sprint several years ago. A graphic produced by towerco Crown Castle, reproduced in Moffett's report, shows substantially better signal propagation for 2.5GHz than C-band spectrum.

But elsewhere, and especially in advanced Asian markets like South Korea, operators have benefited from much denser networks. In mid-2021, Ericsson estimated there were as many as 166,250 5G sites in South Korea, equating to 57 for every 10,000 people. While it is unclear if all these were macro sites, as opposed to small cells, the equivalent figures for Europe and North America do look to be substantially lower. In mid-2018, for example, Deloitte estimated there were just 8.7 mobile sites per 10,000 people in Germany, and only 4.7 in the US. There has been no massive densification wave in these regions since then.

Poor coverage is arguably the biggest impediment to the rollout of "standalone" 5G, the version of the standard accompanied by a new core network and not simply comprising updated radios. UK operators already have the core network technologies in place to support standalone.

But with "non-standalone," its forerunner (which uses the 4G core), they could rely on 4G networks for the uplink signals that devices routinely send, even during idle time for customers. Because standalone removes that 4G uplink option, 5G coverage needs to be widespread before it can be launched at scale. In the UK, as noted, it is not.

Whether the UK's 5G woes are important is another matter. Years ago, it was fashionable for governments and other stakeholders to talk up the standard as an economic spur. Keir Starmer, the UK's recently installed prime minister, is currently on a tirade against the opposition Conservative Party, complaining about the miserable state in which it left UK finances and society. It would not be a total surprise if he blamed his predecessors for the condition of the country's 5G infrastructure, too.

But, there is still nothing a consumer or business can do on a 5G network that cannot be done on 4G, and those 4G networks are good. Unless that changes, customers have little cause for complaint.

Update: Rootmetrics approached Light Reading after this story was published with the following clarification about its methodology: "The point we wanted to make about background testing, is that a 5G signal may be available, but because the device is not in an active state, in many cases the operator itself will keep it connected to 4G rather than 5G – as it clearly doesn't require 5G's capacity. It therefore won't latch onto that 5G signal, and will be recorded as time spent on 4G, lowering the 5G availability figure."