Do You Actually Need 5G? 4G vs 5G and FWA

The short answer, and why it matters

The 4G vs 5G decision is an engineering and commercial one, not a race to buy the newest acronym. The question that matters is not simply whether 5G is faster than 4G in the abstract. It is whether the extra money produces an actual operational benefit for the specific thing you are connecting.

Most of the time the honest answer is that it does not. A vending machine, an alarm panel, a payment terminal or a remote PLC sends a trickle of data and cares far more about being reachable than about peak speed. A weak 5G signal in a metal cabinet will lose to a strong 4G signal on a well placed external antenna every time. The best technology for a ten year deployment is rarely the one with the highest download figure on the day you buy it.

There is a real category of work, though, where 5G is not marketing. When you are pushing a lot of data uphill to the cloud, or using cellular to replace a fixed line at a whole premises, 5G changes what is possible. Fixed wireless access is the clearest example, and it is where most of this guide focuses. The companion piece to this one deals with the low power, low data end of the market, LTE categories in depth, and where 5G RedCap and eRedCap fit for IoT devices, covered in our forthcoming 5G for IoT guide: [URL TO VERIFY].

Where 5G genuinely earns its cost

5G stops being a badge and starts being useful when the application needs sustained throughput, low latency under load, or a lot of upload capacity. In practice, most of those cases in the UK come down to one thing: using cellular as a primary wide area network rather than as a telemetry backhaul.

The largest single use case is fixed wireless access, or FWA. Here a 5G connection delivered to a building, often through an outdoor unit on a wall or pole, replaces a fixed broadband line. There is no trenching, no waiting months for a fibre build, and no landlord permission for ducting. For premises where fibre is unavailable or uneconomic, this is a genuine alternative to a leased line or FTTC, and it is the application that has driven most of the outdoor 5G router market over the last three years. We cover the hardware side of this in detail in our guide to outdoor 5G routers, and the full technical and commercial breakdown of UK FWA lives on our sister reference at 5gfwa.co.uk.

Beyond FWA, 5G earns its keep for multiple HD or 4K CCTV streams, rapid retrieval of footage from vehicles, pop-up offices and construction sites running cellular as the main site connection, broadcast and live production, large software or digital signage updates, edge systems uploading images and models, passenger Wi-Fi on transport, and private 5G on industrial sites where slicing and low latency are contracted rather than assumed. What these share is that they move real volumes of data, or they need guaranteed performance that only a standalone 5G service can offer.

How fast will your connection actually be?

This is the question buyers most often get wrong, because the number on the box is a theoretical peak measured under laboratory conditions. Real throughput depends on signal quality, distance to the mast, how many other users share the cell, the spectrum bands your hardware supports, and how well the antenna is installed. A 5G router with two bars will regularly underperform a 4G Cat 6 router with five bars.

It also helps to remember that 4G is not one technology. It is a family that spans four orders of magnitude in capability, and the right comparison is often not "4G versus 5G" at all but "which LTE category versus RedCap versus full 5G". The table below gives approximate theoretical peaks. Treat every figure as a ceiling you will rarely touch, not a speed you will see.

The practical point is that a customer weighing up a router is often really deciding between Cat 4, Cat 12 and 5G, not between "old" and "new". For a great many industrial jobs, a Cat 4 router already delivers far more capacity than the application will ever use.

5G NSA versus 5G SA: what the 5G symbol does not tell you

Most people see a 5G symbol and assume they have the complete 5G proposition. They usually do not. Much of the public 5G in service is Non-Standalone, or NSA, which bolts a 5G radio layer onto a 4G core. It gives you more bandwidth, which is worthwhile, but the more distinctive 5G capabilities need Standalone, or SA: a full 5G core, better service control, deterministic low latency and network slicing.

UK SA is real and growing rather than finished. Virgin Media O2 has switched on 5G SA across hundreds of towns and cities, EE and Vodafone have SA services, and VodafoneThree has committed, as part of an 11 billion pound programme, to reach 99 percent of the population with 5G Standalone by 2030 and 99.96 percent by 2034, from a baseline near 47 percent at the point of merger. That is a rollout in progress, not a nationwide fact you can assume at any given postcode.

So before you buy a 5G router for its slicing or low latency features, it is worth asking a short list of questions: Is 5G SA available at the installation address? Does the SIM or roaming agreement actually provide access to it? Does the router support the required SA bands? Is slicing sold as part of the service, or merely technically possible? And is the application even designed to benefit? If the honest answers are no, you are buying a 5G router mainly for extra bandwidth, which is a legitimate reason but a different one.

Why uplink matters more than the headline speed

Consumer advertising is built around download speed, because that is what streaming and browsing consume. A large share of demanding IoT and business applications are the other way round. They care about upload: CCTV and machine vision sending footage, vehicles transmitting video, digital signage synchronising, remote backup, and edge systems pushing large datasets to the cloud.

This is why "up to 1 Gbps" tells you almost nothing on its own. The figures that decide whether a deployment works are the expected uplink throughput, the latency and jitter, the data allowance, and how congested the cell is at the times you need it. It is entirely possible for a headline gigabit service to have an uplink that chokes on a handful of simultaneous camera streams. If your workload is upload heavy, that is the number to interrogate, and it is one of the strongest reasons to choose 5G where SA and good coverage line up, because SA uplink can be markedly better than a contended 4G cell.

One site, more resilience: bonding and satellite

Sometimes the answer to "how do I get more out of this connection" is not a faster single link but a smarter combination of links. Two techniques are worth understanding, and both keep 5G at the centre rather than replacing it.

Bonding several SIMs into one session

Bonding pools two or more cellular links into a single logical connection that survives any one link dropping. This is not failover, which switches after a break, and not load balancing, which splits separate sessions across links. Bonding holds one session up across all of them, which is the behaviour that matters for live video and always on uplinks. On Teltonika hardware this is now a one click install, and a dual modem unit can bond two live 5G networks, or a pair of units can bond four SIMs across four networks. We walk through a real build in bonding SIMs with Bondix on the RUTM52, including where the honest limits sit.

Adding satellite as a second transport

5G is at its strongest when it is not working alone. The most resilient site designs pair it with a second transport that fails in a different way, and increasingly that means low earth orbit satellite. 5G FWA gives low latency and high throughput where there is a mast and clear line of sight, but nothing where there is no mast. LEO gives connectivity almost anywhere with a clear sky, but is more exposed to obstruction and weather. Put them together under an SD-WAN and you get true path diversity: the router steers traffic across whichever link is healthiest and holds sessions up during a failover. In a town edge or industrial estate location, 5G is usually primary and satellite is the always there backup; on a genuinely remote site the roles reverse. We go deeper in our guide to LEO satellite for UK business and IoT. There is also a separate, quieter satellite story for connecting individual low power devices directly, which we cover in our look at satellite direct to device and NB-IoT over NTN.

The unlimited data myth and fair use

5G FWA is usually sold on an unlimited data SIM, because a business premises can easily run into the terabyte range each month, well beyond ordinary IoT tariffs. This is where a common and expensive assumption creeps in: that "unlimited" means unlimited high speed data with no strings. It rarely does.

The lesson is not that unlimited plans are bad, but that the tariff is part of the engineering. Faster connectivity also makes it far easier for an application to consume an unexpectedly large amount of data, which matters a great deal when you are deploying hundreds or thousands of routers. And a cheap consumer unlimited SIM is a fragile foundation for anything you need to depend on: as we document in the case of Three's 3internet APN and CGNAT, a trick that worked for years can quietly stop working as networks change. For a deployment, buy the service you can rely on, not the cheapest SIM that happens to work today.

How to get a public IP on 5G

If you need to reach a device from the outside, to view a camera, connect to a PLC or open an SSH session, you need a routable address. Most consumer and many business SIMs sit behind carrier grade NAT, so the device can talk out but nothing can reach in. There are two clean routes to solve this, plus a do it yourself option for small estates.

Route one: a high data public IP SIM

Specialist IoT connectivity providers sell SIMs that hand you a fixed public or fixed private address that is preserved across network changes, usually with a management portal for the fleet. For an FWA style workload you want one with a large allowance. As a rough market guide in late 2026, high data public IP SIMs are commonly priced around 85 pounds a month and often capped near 500 GB, though pricing and caps vary widely by provider and change frequently, so treat that as an order of magnitude rather than a quote. Our reference on what a public IP SIM is covers the buying questions in detail.

Route two: a breakout or managed VPN service

Rather than exposing the device on a public address, a breakout service terminates the SIM traffic into the provider's network and hands you a private, routable endpoint over a VPN, often with a fixed IP behind it. This is the cleaner drop in replacement for the old port forwarding model, because the addressing is predictable and does not change when masts or policies do. For fleets it also centralises firewalling and allow listing.

Route three: your own VPS and tunnel

If you have the skills and only a site or two, you can solve carrier grade NAT yourself with a small cloud VPS holding a static public address and a WireGuard tunnel back to the site, giving you a fixed endpoint for a few pounds a month. It is a genuine alternative for a technical team, though not a substitute for an SLA and a support desk once you are running an estate. We show ten practical versions of this in our piece on using a low cost VPS for IoT and M2M.

Coverage, antennas and the real cost of 5G

A 5G router cannot repair a poor installation. In many cases a properly installed 4G router with an external directional antenna will beat an expensive 5G router left inside a metal cabinet. Mid band 5G around 3.5 GHz, which carries most of the capacity, is far more sensitive to distance and line of sight than 4G at 800 MHz, so cell proximity and antenna placement matter more, not less, as you move up the technology stack.

The cost comparison also has to look past the router price. Moving to 5G typically means four cellular antennas rather than two, more installation time and antenna separation, higher power draw and heat, a larger data tariff, and often more expensive replacement stock. Multiply any of that across a fleet and the total cost of ownership can look very different from the headline hardware saving. The sensible frame is total deployed cost per site over the deployment life, not the price on the box.

Where 4G is still the right choice

For a large share of connected estates, 4G is not a compromise but the correct engineering answer. Vending machines, ATMs and payment terminals, alarm panels, energy monitoring, building management, PLC access, basic fleet telemetry, agricultural monitoring, digital signage with scheduled downloads, failover for a modest business line, remote equipment support, and low resolution or event triggered CCTV all live comfortably within LTE Cat 1 to Cat 4.

In these cases the discipline is to resist buying capability the application will never use. A Cat 4 router with a good antenna and a sensible tariff will be cheaper to buy, cheaper to run, easier to power, and no less reliable than a 5G unit, often more so, because it is not chasing a mid band signal that comes and goes. Reliability, coverage and uplink almost always matter more than peak download for this class of device.

Is it safe to buy 4G in 2026?

For the record: all UK mobile network operators have completed their 3G switch off. 2G is a longer story and is not imminent. Operators have committed to switch off 2G by 2033 at the latest, with EE starting from May 2029, O2 from summer 2029 and Vodafone during 2030, while Three has never run a 2G network. Government and industry agreed a 2G Switch-Off Charter in March 2026 to manage the process. None of that points at 4G being withdrawn in the foreseeable future.

No responsible supplier should promise that LTE will run forever. The sensible question is whether the device, its modem and bands, the operator agreement and the manufacturer support will comfortably cover the expected deployment life. For a router bought in 2026, the answer for LTE is very likely yes.

Who offers what in the UK

All four UK networks, EE, Vodafone, Three and O2, now offer 5G and 5G business or broadband products, though Vodafone and Three now sit under the single VodafoneThree operator following their 2025 merger. Their strengths differ in ways that matter for a deployment, and the picture below is a snapshot for late 2026 rather than a fixed ranking, because coverage and tariffs move constantly.

EE has the largest 5G geographic footprint, helped by 700 MHz low band deployed under the Shared Rural Network, which makes it the natural choice for rural and semi rural sites; for FWA it is generally a data SIM plus your own hardware rather than a packaged home product. Three holds the largest mid band spectrum block and posts the fastest average 5G speeds, with a dedicated home FWA product, but weaker indoor penetration. Vodafone pairs aggressive mid band rollout with a leading position on Standalone 5G and slicing, and a 5G Broadband product plus a strong business SIM portfolio with SLA support. O2, part of Virgin Media O2, has broad urban 5G and an early, wide 5G SA switch on, but no prominent consumer FWA product, so it is typically accessed for FWA through data only SIMs. For a full head to head aimed at choosing a network by location, see the UK 5G FWA network breakdown.

A 4G vs 5G decision framework

Reduced to its essentials, the choice usually starts from the workload and the site, not the technology. The table gives a sensible starting point, which you then temper with coverage, uplink and cost at the actual address.

Whatever the starting point, a handful of questions will save most of the expensive mistakes. What does the application actually send, and how much of it is upload? Is there strong signal at the exact install point, and can the antenna be placed well? Is SA, slicing or a public IP genuinely available and contracted, not just theoretically possible? What is the realistic data consumption across the fleet, and does the tariff and its fair use policy support it? And will the hardware, bands and support cover the full deployment life? Answer those honestly and the 4G or 5G decision usually makes itself.

Common questions