4G Sunset UK

Network Generations

4G Sunset UK: When Britain’s Workhorse Network Actually Retires, and What It Means for IoT

3G has gone and 2G is on a countdown to 2033. So it is reasonable to ask which generation is next. The honest answer for 4G surprises people, and it reshapes how anyone deploying connected devices should be thinking about the years ahead.

Every conversation about mobile network sunsets in the UK eventually lands on the same nervous question: if 3G is already switched off and 2G has a firm deadline, when does 4G go the same way? For anyone running a fleet of connected devices, the answer is not academic. A sunset date dictates procurement cycles, firmware roadmaps, warranty commitments and the total cost of a deployment that might need to stay in the field for a decade.

The short version is that 4G is not sunsetting any time soon, and no UK operator has announced a date. But that is not the end of the story. The strategic direction of travel is already clear, and it runs through 5G Standalone, network slicing and a family of reduced-capability standards designed to carry exactly the kind of traffic that 4G handles today. This article walks through where the sunset actually stands, why 4G is different, and how the successor technologies change the calculation for IoT.

Where the UK mobile sunset actually stands

It helps to separate what has happened, what is committed, and what is merely speculated. The picture across the four mobile network operators (EE, Virgin Media O2, Vodafone and Three) breaks down cleanly by generation.

Generation	Introduced	Status in the UK	Relevance to IoT
2G (GSM)	1991	Committed off by 2033, several operators starting sooner	Still the fallback bearer for alarms, telecare and many smart meters
3G (UMTS)	2003	Effectively switched off across all operators	No new deployments; legacy devices must be migrated
4G (LTE)	2012	No announced sunset date; still expanding	The current workhorse for the majority of cellular IoT
5G (NR)	2019	Rolling out; Standalone cores maturing	The strategic home for future IoT via RedCap and slicing

For the operator-by-operator detail behind these headlines, and the fuller argument on timing, see the companion analysis on when will 4G be switched off in the UK. The summary below focuses on what each stage means for connected-device estates.

3G: the switch-off that has already bitten

The 3G shutdown is the one that has already caused real pain in the field. All four UK operators have retired their 3G networks, freeing the spectrum for more efficient 4G and 5G services. For consumers the impact was designed to be negligible, because almost everyone already carries a 4G or 5G handset. For IoT, it was a different experience entirely.

The devices caught out by 3G retirement were rarely the newest. They were the routers, trackers, vending telemetry units and monitoring gateways installed years earlier, quietly working, and never revisited. Many were 3G modules with a 2G fallback, which meant that when 3G disappeared they dropped to a slower, older bearer rather than failing outright. That masked the problem in some estates and delayed the reckoning. Others had no fallback and simply went dark.

The specific casualties were instructive. Older payment terminals and cash machines that dialled home over 3G needed swapping or reconfiguring. Vehicle telematics and stolen-vehicle tracking units, many fitted years earlier and never touched since, lost their primary uplink. First-generation connected CCTV and DVR backhaul units, remote environmental loggers and vending machine telemetry all featured on the casualty list. The common thread was that these were unattended machines with no user to notice a warning, frequently running on wholesale or roaming SIMs where the network had no direct line to the end operator. Working out which SIM sat in which asset, and who owned the decision to replace it, turned out to be the hardest part, and in many estates that reconciliation work is still ongoing.

The lesson that carried into the wider sunset conversation was uncomfortable but simple. A connected device is only as future-proof as the radio inside it, and the radio is the component most often forgotten once a deployment is signed off. That lesson is the reason the 4G question matters so much now: nobody wants to repeat the 3G scramble on a larger scale.

2G: the long goodbye running to 2033

If 3G is the past, 2G is the slow-motion present. In December 2021 the government and the mobile industry agreed that 2G and 3G would both be retired by the end of 2033 at the latest. In March 2026 the government went further, publishing a voluntary 2G switch-off charter setting out how operators would support a smooth transition. The individual operator timelines are staggered.

Operator	2G position
EE	Has signalled it will begin switching off 2G from around 2029
Vodafone (VodafoneThree)	Plans to switch off its 2G network during 2030
Virgin Media O2	Began moving traffic away from 2G from 2025, with 2G retained longest for certain services
Three	Never operated a 2G network; launched as a 3G-only operator

The reason 2G has outlived 3G, despite being the older technology, is that it became the connectivity of last resort for a huge installed base of low-bandwidth machines. Smart energy meters in parts of southern England and Wales use 2G and 3G to talk to suppliers. Telecare devices such as pendant alarms, fall detectors and sensors rely on it, often as a backup path. Basic alarm panels, agricultural sensors and older trackers lean on 2G because it is cheap, power-frugal and almost universally present.

That deep dependence is exactly why the deadline stretches to 2033. Unlike a phone, these devices have no user to prompt an upgrade, no screen to display a warning, and frequently no direct relationship with the network at all because they run on roaming or wholesale SIMs. Identifying and replacing them is slow, manual work. For any organisation still running 2G-dependent hardware, the switch-off charter is a signal to start auditing now rather than in 2032.

So when does 4G actually switch off?

Here is where expectations and reality diverge. There is no announced date for a UK 4G switch-off. None of the operators has published one, the government has not set a target, and there is no regulatory deadline of the kind that governs 2G and 3G. When people talk about a 4G sunset, they are describing a trajectory, not a scheduled event.

The reasons are structural. 4G is not a legacy niche; it is the mainstream. As of recent Ofcom reporting, 99% of UK premises can receive a 4G signal indoors from at least one operator, and coverage is still being extended into rural not-spots under the Shared Rural Network programme, with a target of 95% of UK landmass. A network that is still being built out is not a network anyone is preparing to dismantle.

4G also underpins 5G itself in most of the country. The dominant deployment model to date has been 5G Non-Standalone, in which the 5G radio rides on top of a 4G core and control plane. In that architecture, turning off 4G would break 5G. Until Standalone 5G is widespread, 4G is load-bearing in the most literal sense.

The realistic outlook, then, is that 4G persists well into the 2030s and quite possibly beyond, long after 2G and 3G are memories. It will not be switched off on a date. It will be gradually thinned, refarmed and eventually superseded as traffic migrates to 5G, in a process measured in years and driven by economics rather than a headline announcement.

Why 4G remains the IoT workhorse

For connected devices specifically, 4G is not a compromise while everyone waits for 5G. It is the pragmatic default, and it earns that position on several fronts.

It covers the whole IoT performance range

The 4G family spans a remarkable spread of requirements. At the low-power end, LTE-M and NB-IoT carry small, infrequent messages from battery-powered sensors, meters and trackers, with years of battery life and deep indoor penetration. In the middle sit the LTE device categories, Cat-1 and Cat-4 in particular, that serve routers, payment terminals, cameras and gateways needing steady throughput. One technology generation covers everything from a soil-moisture probe to a full 4K surveillance feed.

It suits long device lifecycles

Industrial and infrastructure IoT is deployed to last. A connected water-quality monitor, a traffic controller or a remote pumping-station gateway may be expected to run for ten to fifteen years. 4G modules are mature, cheap, widely certified and supported by a deep ecosystem of vendors. For a device being specified today with a decade-long horizon, 4G with a clear onward migration path is a defensible engineering choice in a way that 3G stopped being years ago.

It is available now, everywhere

5G coverage is growing quickly but is still concentrated in populated areas, and Standalone 5G more so. 4G is effectively ubiquitous. For a national deployment that must work in a rural depot as reliably as a city centre, that ubiquity is decisive. The workhorse label is earned precisely because 4G turns up and does the job wherever the job happens to be.

What will eventually bring the 4G sunset closer

If there is no date, what actually drives the eventual retirement of 4G? Three forces, all of them gradual.

The first is spectrum economics. Every band an operator dedicates to 4G is a band it cannot use for 5G, which is far more spectrally efficient. As 5G traffic grows, the pressure to refarm 4G spectrum increases. This is already visible at the edges, with slices of legacy spectrum being repurposed generation by generation.

The second is energy and operating cost. Running parallel network generations is expensive and power-hungry. Operators have strong incentives to collapse the number of technologies they maintain. 5G is markedly more energy-efficient per bit than older generations, and consolidating onto it lowers running costs, which is part of why 2G and 3G were targeted first.

The third is the maturity of the successor. 4G cannot be meaningfully wound down until 5G Standalone can carry not just consumer broadband but the full breadth of IoT traffic that 4G handles today, from the smallest sensor to the busiest gateway. That capability is the missing piece, and it is exactly what the next standards are designed to supply.

Enter 5G Standalone: the foundation for everything next

Most 5G in the UK so far has been Non-Standalone, which delivers faster speeds but still depends on a 4G core underneath. 5G Standalone (SA) is the full architecture: a native, cloud-native 5G core with no 4G dependency. The distinction sounds like plumbing, but it is the enabler for almost every feature that matters to IoT.

Standalone brings lower latency, more responsive connection setup and, crucially, the control that lets an operator treat different traffic types differently on the same physical network. It is the platform on which network slicing, guaranteed service levels and reduced-capability device types all depend. Without SA, 5G is essentially a faster pipe. With SA, it becomes a programmable connectivity fabric. That is the shift that eventually makes retiring 4G thinkable, because SA can host the IoT workloads that currently have nowhere else to go.

Network slicing, explained without the jargon

Network slicing is one of those terms that gets used constantly and explained rarely. The idea is straightforward. A slice is a dedicated, logically separate virtual network running over the same shared physical infrastructure, configured with its own performance characteristics.

Picture a single motorway. Today, all traffic shares the same lanes, so a surge of one type of vehicle slows everything down. Slicing is the equivalent of permanently reserving lanes tuned for specific journeys: one lane guaranteed for emergency vehicles that must never be delayed, another optimised for slow, heavy freight, another for high-speed through-traffic. Each lane behaves as though it were a private road, even though the tarmac is shared.

For IoT the implications are significant. A utility could run a slice for critical grid telemetry with guaranteed low latency and high reliability, entirely insulated from a separate slice handling bulk, non-urgent meter reads. A hospital could isolate connected medical devices from guest and administrative traffic on the same network. A critical alarm signal and a large firmware update would never again compete for the same capacity, because they would live in different slices with different guarantees.

Slicing turns a mobile network from a single best-effort service into a portfolio of tailored services. For applications where reliability is non-negotiable, that is a genuine step change, and it is only possible on a Standalone 5G core.

RedCap and eRedCap: serving IoT natively on 5G spectrum

The final piece of the puzzle answers an obvious objection. Full 5G modems are powerful, expensive and power-hungry, wildly over-specified for a sensor that sends a few kilobytes a day. If 4G eventually retires, what carries the vast middle and lower tiers of IoT that never needed full 5G in the first place? The answer is a deliberately stripped-down family of 5G device types.

RedCap, short for Reduced Capability and defined in 3GPP Release 17 (frozen in 2022), takes full 5G New Radio and removes what most IoT devices do not need: it narrows bandwidth to around 20 MHz, cuts antenna counts and drops features like carrier aggregation. The result is a device delivering roughly 150 Mbps peak downlink at a fraction of the cost and power of a full 5G modem. That comfortably covers industrial sensors, wearables, video surveillance and mid-tier gateways.

eRedCap, or enhanced RedCap, arrived in 3GPP Release 18 (frozen in June 2024) and pushes further down. It halves the channel bandwidth to 5 MHz, restricts the device to a single receive antenna, adds half-duplex operation and caps the peak rate at around 10 Mbps. The target is not wearables but the enormous volume of devices currently served by LTE Cat-1: basic sensors, meters and trackers that need reliable, low-cost connectivity rather than speed. Detailed technical breakdowns of the Release 18 device type are maintained at eredcap.com.

Crucially, both are built on the 5G Standalone architecture, which means they inherit slicing, better power-saving modes and native 5G core integration. This is the mechanism by which IoT is served natively on 5G spectrum rather than being marooned on ageing 4G. Where these technologies sit relative to the older cellular IoT standards is easiest to see side by side.

Technology	3GPP era	Typical peak rate	Best-fit IoT use
NB-IoT	4G LPWAN	Tens of kbps	Static meters, deep-indoor sensors
LTE-M	4G LPWAN	~1 Mbps	Mobile trackers, low-rate sensors with voice
eRedCap	5G, Release 18	~10 Mbps	Cat-1 replacement: meters, basic sensors, trackers
RedCap	5G, Release 17	~150 Mbps	Wearables, industrial sensors, surveillance
Full 5G NR	5G eMBB	Gbps class	FWA, high-bandwidth video, dense broadband

A note on what comes after: 3GPP Release 19, frozen in December 2025, did not introduce a further reduced-capability device type. Instead it extended RedCap toward non-terrestrial (satellite) networks, added wake-up receiver work for even better battery endurance, and standardised an initial Ambient IoT device type as a separate paradigm. In other words, the near-term IoT roadmap on 5G is RedCap and eRedCap, with satellite and ambient concepts opening up behind them.

Where the UK actually is with RedCap and 5G SA

This is not purely theoretical. EE was first to announce commercial RedCap deployment in the UK, with coverage rolling out across major cities from the middle of 2025. Vodafone and Three have confirmed their own RedCap roadmaps as their Standalone footprints expand. Commercial RedCap modules from the major cellular vendors began appearing in 2024, and eRedCap silicon is following, with modules expected to reach industrial and automotive form factors first before consumer packages arrive as volumes build.

The trajectory is clear even if the timeline is gradual: Standalone coverage widens, RedCap becomes a routine option alongside 4G, eRedCap matures into the natural Cat-1 replacement, and the centre of gravity for new IoT design slowly shifts from 4G onto 5G. None of this forces an immediate change, but all of it should inform decisions being made today.

Private and campus 5G: slicing brought on site

One of the more consequential developments for industrial IoT sits slightly outside the public-network sunset story: private 5G. Where a factory, port, logistics hub, mine or large campus needs guaranteed connectivity across a defined area, a dedicated on-premises 5G Standalone network delivers the same slicing, low latency and reliability benefits as the public network, but under the site operator’s own control.

The appeal for heavy industry is concrete. A private network keeps sensitive operational data on site rather than traversing a public core. It provides deterministic performance for automated guided vehicles, robotics, machine-vision quality inspection and safety systems that cannot tolerate variable latency. And because it is built on Standalone architecture, it supports the same RedCap and eRedCap device types, so the connected sensors, cameras and handhelds across a site can be right-sized to their traffic just as they would on a public network.

This matters to the 4G question because private 5G gives industrial adopters a reason to move to 5G Standalone on their own timetable, independent of when public operators refarm their spectrum. A manufacturer commissioning a new line today can specify a private SA network with slicing from day one, deploying RedCap devices natively rather than defaulting to 4G. Every deployment of this kind quietly shifts more of the IoT installed base onto 5G and, over time, thins the population of devices that depend on 4G at all. Private networks are not a mass-market phenomenon, but in exactly the industrial settings where long-life connected infrastructure concentrates, they are an accelerant for the transition.

A practical migration playbook

For anyone responsible for connected devices, the sunset picture translates into a handful of concrete actions.

• Audit the radio, not just the device. Know exactly which bearer every device in the estate depends on, and flag anything still relying on 2G or a 2G fallback. Those are the units on a clock.
• Treat 4G as a sound primary bearer, with conditions. For deployments over the next few years, 4G remains sensible. Specify modules that are certified across LTE and, where the budget allows, capable of RedCap or a clear upgrade path.
• Match the technology to the traffic. Do not over-specify. A sensor sending a daily reading wants NB-IoT or eRedCap, not full 5G. A surveillance gateway wants RedCap or LTE Cat-4. Right-sizing controls both cost and power.
• Plan around device lifecycle. A device with a fifteen-year life needs a fifteen-year connectivity plan. If it cannot be upgraded in the field, it must be specified for a bearer expected to survive the deployment, which today means LTE plus a credible 5G path.
• Watch Standalone coverage in your operating areas. RedCap and slicing depend on 5G SA. Where you deploy determines when these options become real for you, so track your operators’ SA footprint against your geography.

Frequently asked questions

Is 4G being switched off in the UK?

Not in the way 2G and 3G are. No UK operator has announced a 4G switch-off date, there is no government deadline, and 4G coverage is still being expanded. 4G will be gradually thinned and refarmed as traffic moves to 5G, but this is a slow process measured in years, not a scheduled event.

When will 4G actually be switched off?

There is no set date. The realistic expectation is that 4G persists well into the 2030s and probably beyond, long after 2G and 3G are gone. It also currently underpins most 5G in the country through Non-Standalone deployment, so it cannot be retired until Standalone 5G is widespread.

Should I still deploy 4G IoT devices?

For most deployments over the next several years, yes. 4G remains the pragmatic default bearer with the widest coverage and the deepest device ecosystem. The important caveat is to specify hardware with a migration path, ideally modules capable of RedCap or with a clear upgrade route, rather than 4G-only devices intended to last fifteen years.

What replaces 4G for IoT?

Not a single technology, but a combination: 5G Standalone as the foundation, network slicing for guaranteed service levels, and the RedCap and eRedCap device types carrying the mid- and lower-tier IoT traffic that 4G handles today, all served natively on 5G spectrum.

This article is independent analysis from Peter Green, a Cellular Connectivity Specialist – My first mobile phone was in 1995 and was an NEC P100 on the BT Cellnet network 🙂