London's mobile networks are running out of room

Ookla Research™ Articles

London has slipped from seventh to last among 18 European capitals for median mobile download speed since 2021.

London packs more jobs and people into its core than any other UK city, and its mobile networks have not kept pace with that demand. An early start on 5G lifted London to seventh of 18 European capitals for median download speed in 2021, but traffic has since grown faster than capacity, and adding capacity in central London has become harder. Using Speedtest Intelligence® data from 2018 to Q3 2026, alongside Ookla background signal measurements and official statistics, we compare London with Europe’s other capitals and the UK’s largest cities, show where its networks strain, and examine why new capacity is so hard to add.

Key Takeaways:

• London has fallen from seventh to last among European capitals on median mobile download speed. Its median download speed rose almost sevenfold from Q1 2018 to 109.43 Mbps in Q3 2026, but other capitals improved faster. Since early 2023, it has also had the slowest upload speed of the 18 every quarter, and the highest latency under load for the worst 10% of samples.
• 5G adoption has outrun 5G capacity. London’s median 5G download speed fell from 200 Mbps in Q2 2021 to 133 Mbps in Q3 2026 as the share of tests on 5G nearly doubled to 80%, leaving it with the slowest 5G median of the 18 capitals. Paris, a comparably large capital, has held its 5G median above 225 Mbps since mid-2021.
• London’s networks congest as commuters arrive. In the City of London and Westminster, weekday lunchtime download speeds run 27% slower than at weekends, with 26% higher latency under load. London still has the lowest mobile latency of the 15 UK cities we studied, helped by its place at the center of the UK’s internet.
• Adding capacity is getting harder. Applications for new street masts in London have fallen by more than four-fifths since 2022, almost three in four decided since 2019 were refused, and 5G phones in neighborhoods fully covered by conservation areas spend about 4 percentage points less time on 5G.
• London’s least affluent neighborhoods are 18% faster than its most affluent. Within the same borough, about a third of that gap comes from phones spending more time on 5G and about a sixth from denser populations nearby.

London’s operators have spent the past five years building 5G while demand kept climbing, and UK mobile data use rose 18% in 2025 alone, Ofcom reports. VodafoneThree, formed in May 2025, is now combining two networks, and the Mayor’s draft London Plan, in consultation until 15 October 2026, sets out new expectations on mobile connectivity. Behind both sits a long dispute over the rooftops and land that London’s networks depend on, which we return to below.

Most of Europe’s capitals have overtaken London

London was among the first European capitals to launch 5G, which lifted it to seventh of 18 for median download speed by mid-2021. By Q3 2026, however, it had the slowest median download speed of the 18, at 109.43 Mbps, behind Rome (125.26 Mbps) and well below the 172.22 Mbps median of the other capitals. It also had the slowest median upload speed (11.40 Mbps) and the highest latency under load for the worst 10% of samples (2,204 ms). Latency is the time data takes to make a round trip across the network, and latency under load measures it while the connection is busy with a large download, showing how a network copes when it is in heavy use.

The decline tracks 5G. London’s 5G median peaked at 200 Mbps in Q2 2021, when 41% of tests ran on 5G, fell to 102 Mbps by Q3 2023 and has recovered only to 133 Mbps, with 80% of tests now on 5G. Phones moved onto 5G in London at least as fast as in Paris or the rest of the UK, but capacity did not keep up, and London’s 5G median is now the slowest of the 18 capitals. Across the UK, 5G medians fell in 12 of 15 large cities between Q3 2022 and Q3 2026 as 5G use rose in all of them, though Manchester’s climbed from 137 to 178 Mbps.

Within the UK, London features one natural advantage. Its median multi-server latency (measured against several nearby test servers on an idle connection) of 38 ms is the lowest of the 15 UK cities we studied, and latency rises about 3 ms for every 100 km from the capital, to 61 ms in Belfast. Operators like Three and BT’s EE have moved parts of their core networks out to regional sites, but much traffic is still hauled back towards London, where most UK interconnection sits. For example, LINX’s main London exchange peaks at nearly 9.8 Tbps, against about 0.9 Tbps at LINX Manchester. Speed shows no such pattern, and London’s median download ranks only eighth of the 15 UK cities.

London’s mobile slowdown arrives with the commuters

The City of London and Westminster hold about 1.5 million jobs but only 213,000 residents. On weekdays, their median download speed falls through the morning to 80.07 Mbps at 13:00, 27% below the same hours at weekends, while latency under load rises 26%. This is a classic case of network congestion. As commuters arrive, more phones share the same cells, and each gets a smaller slice of their capacity. Ookla data on YouTube start times shows the same squeeze.

With offices largely empty in 2020 and 2021, as Londoners worked from home, the City had the fastest mobile speeds of London’s 33 boroughs, and in 2026 the slowest. The reversal reflects commuters returning to a district with few residents and dense office blocks, where extra rooftop capacity is hardest to add.

The squeeze varies by operator. Over the past year, EE customers saw the smallest weekday lunchtime slowdown in central London (15%) and Three customers the highest median download speed across London (113.27 Mbps). O2’s median latency under load of 1,573 ms is about three times its rivals’, but the gap stays as wide overnight, when networks are quietest, which points to how its network queues traffic during downloads (known as bufferbloat) as well as to congestion.

Why London’s least affluent neighborhoods are faster

The commuter belt tells a different story. Median download speed ranges from 122.57 Mbps in Barking and Dagenham, one of London’s least affluent boroughs, to 89.01 Mbps in the City of London. Ranking London’s 4,994 census neighborhoods by England’s official Index of Multiple Deprivation, the least affluent fifth records 107.27 Mbps against 96.01 Mbps in the most affluent. They are more than twice as densely populated, have about two-fifths less tree cover and half as much land in conservation areas, and their phones run 78% of tests on 5G against 74%.

Comparing neighborhoods within the same borough, about a third of the 18% speed lead comes from phones spending more time on 5G and about a sixth from denser populations nearby, where demand may draw more investment. Tree cover and operator mix each explain only a small share, while flats and built-up land work the other way, since signal struggles to reach inside dense blocks. Our data cannot cleanly show where operators are able to secure sites, which may also differ between richer and poorer areas.

Adding capacity in central London is getting harder

In the UK, mobile operators own few of the sites their antennas use. They rent space on towers and rooftops, from tower companies or directly from landowners, and even where a tower company runs a site, the roof or land beneath it usually belongs to someone else. In central London, where there is little room for ground-based masts, much of the network sits on rooftops.

The terms are set by the Electronic Communications Code, and its December 2017 reform cut what landowners earn. Rents are now set on a no-network valuation, as if no mobile operator wanted the site, and a 2019 tribunal ruling valued a London council rooftop at £1,000 a year, against £21,000 under the old regime. Since April 2026, the same approach has applied to renewals of older leases. Building owners now have less reason to host antennas, and many may look to rooftop solar or gardens instead. Landlords can also end a site lease with a notice to quit, typically when a building is redeveloped. Mobile UK estimates that notices to quit keep around 7% of London’s rooftop sites out of action at any time, and VodafoneThree expects to take nearly 100 sites offline by the end of 2026, with replacements taking around five years. Landowner groups counter that the reforms cut some rents by up to 90%.

Unlike France, where the ANFR register shows 4G or 5G sites within 10 km of central Paris up 26% from August 2022 to August 2026, the UK has no public register of mobile sites. London’s planning records show the pipeline instead. Applications for new street masts peaked in 2020, early in the first 5G build, and have fallen by more than four-fifths since 2022, while applications to upgrade existing sites fell by only about a fifth over 2022 to 2025. Almost three in four new street masts decided since 2019 were refused, against one in five upgrades. As a mitigation, operators are adding small cells, low-powered radios on lampposts and street furniture, including about 3,400 on Transport for London’s roadside assets. The loss of rooftop sites has acted as a forcing function for this kind of street-level densification, with London now well ahead of many European cities on the depth and breadth of small cell deployment.

Conservation areas, which protect places of special architectural or historic interest and where planning rules restrict telecoms equipment, cover 77% of Westminster. Within boroughs, new masts inside them are 9 to 10 points more likely to be refused, and 5G-capable phones in neighborhoods fully covered by them spend about 4 percentage points less time on 5G than comparable neighborhoods, against about 83% where there is no conservation area. The gap is 7 points for EE and 6 for Three customers.

A benchmark for London and other global cities

This analysis builds on our joint report with the Greater London Authority, A New Lens on the Connected Capital, whose Capital Connectivity Standard is designed as a blueprint for other cities. The Mayor’s draft London Plan, due for examination next year, now states that notices to quit “have eroded London’s overall mobile connectivity compared to other equivalent cities”. It asks developers to keep mobile service running through construction and to plan masts on taller buildings and in the public realm, and it asks boroughs to work with operators on sites in conservation areas.

Ookla retains ownership of this article including all of the intellectual property rights, data, content graphs and analysis. This article may not be quoted, reproduced, distributed or published for any commercial purpose without prior consent. Members of the press and others using the findings in this article for non-commercial purposes are welcome to publicly share and link to report information with attribution to Ookla.