Nobody Has Told Anyone When 4G Dies. That’s RedCap’s Real Problem.

Omdia has stepped back from RedCap and put its weight behind eRedCap. The direction is right. But its own data contains the proof of what actually moves this market, and it isn’t module price.

Published 16 July 2026  |  Standards position current at publication

Omdia published a report this month called “5G RedCap will struggle but eRedCap holds promise”. The title does the work.

Three things are holding RedCap back, according to the research: limited 5G Standalone availability, high modem costs, and limited urgency around device longevity. RedCap will see some adoption but is unlikely to serve an IoT market that did well on LTE and has since had no obvious road forward. eRedCap gets the better hand: better network readiness, a larger device ecosystem, more enterprise urgency.

Alexander Thompson, the senior IoT analyst who led the report, points at module pricing, noting it was “expected to have reduced quickly” and wasn’t.

Most of the coverage will stop there, at the price. The price is the symptom.

Two details that explain everything

In the section on pricing and timing sit the two findings that actually matter, and they’re more interesting than the headline.

The first: many IoT customers would rather buy full 5G modules to future-proof their deployments.

Read that twice if you sell this stuff for a living, because it’s quietly devastating. The entire premise of RedCap is that most devices don’t need full 5G, so why pay for it. The market’s answer has been: because I am putting this box on a wall for ten years and I don’t want to be wrong.

So the buyer overspecifies. Deliberately. The mid-tier product loses the sale not to a cheaper option but to a more expensive one, on the grounds of longevity. Which keeps RedCap volumes low. Which keeps RedCap module prices high. Which makes the full 5G module look like better value per year of service life. Round and round it goes.

The second: there are no immediate LTE switch-off announcements, so urgency stays limited.

This is the one. This is the whole market in a sentence.

Much of the long-term commercial case for RedCap and eRedCap assumes LTE eventually gets refarmed. That’s the argument. It’s what the vendors say, what the analysts say, what this site says. And it’s reasonable.

But almost nobody has put a date on it. Operators have been discussing refarming for years. The public analyst framing has 4G phasing out “beyond 2030”, which is a direction of travel rather than a deadline. A handful of markets have local sunsets in train. None of it adds up to a date a buyer could plan a ten-year estate around.

Without a date, there’s no deadline. Without a deadline, there’s no urgency. Without urgency there are no volumes, and without volumes the module price never falls to the point where the business case closes.

The price problem everyone is discussing is downstream of an announcement nobody has made.

The control experiment

Here’s the part that turns that from an opinion into something closer to evidence, and it’s sitting in the same body of research.

Ask what is moving buyers right now.

In Europe, it’s the Cyber Resilience Act. Secure-by-design obligations and five-year vulnerability patching duties are pulling enterprise buyers toward eSIM and eUICC architectures, according to Omdia’s own analysis. Not because eUICC got cheaper. Because the CRA has teeth and a date.

In North America, it’s the Infrastructure Investment and Jobs Act and the Inflation Reduction Act, driving smart grid, utility and EV charging deployments. Again: not a technology story. Money with timelines attached.

So in the same market, in the same decade, with the same buyers, the instruments carrying deadlines are moving people and the technology transition isn’t. Every argument about RedCap module pricing is happening next to a live demonstration that this industry moves when someone names a date and doesn’t when nobody does.

That’s not a coincidence. That’s the mechanism.

We have watched this work twice

The obvious retort to all of the above is that it’s just logic. Nice chain of reasoning, no evidence.

Except we’ve run this experiment before. Twice.

Operators announced 2G shutdowns. Migration happened. Operators announced 3G shutdowns, with dates, and the M2M industry that had spent fifteen years insisting it couldn’t possibly move got on with moving. Painfully, expensively, late, and complaining. But it moved, and it moved because somebody named a day.

It isn’t only cellular, and the cleanest example we have is the one on our own doorstep. Openreach set 31 December 2025 for the PSTN switch-off, then pushed it to 31 January 2027. Thirteen extra months. The deadline moved. And Ofcom’s figures show PSTN landline customers falling from 5.2 million in July 2024 to 3.2 million in July 2025 anyway. Two million people migrated in the year after the date slipped.

The date doesn’t even have to hold. It only has to exist.

That’s the precedent. Deadlines transition this industry. Nothing else reliably does. Not better technology, not lower prices, not a decade of analyst slides explaining that the new thing is superior.

So the case for an LTE date isn’t speculation. It’s the only thing that has ever worked here.

Which raises the question nobody in the RedCap conversation seems to want to ask.

Why hasn’t anyone named the date?

Not laziness. The answer is structural, and once you see it the whole eRedCap timeline looks different.

LTE is 5G. Not as a marketing line. Formally. When 3GPP made its IMT-2020 submission to the ITU, it submitted NR for eMBB and URLLC, and it submitted E-UTRA/LTE, including NB-IoT and eMTC, as a component radio interface technology in the same package. NB-IoT and LTE-M use the 4G LTE air interface and were accepted as meeting the IMT-2020 requirements for massive machine type communications. 3GPP went further and agreed that LPWA use cases would continue to be addressed only by evolving LTE-M and NB-IoT, with no NR-based LPWA solution studied or specified for the foreseeable future.

Read that again with a switch-off date in mind. Every standardised LPWA technology inside 5G runs on LTE radio, and 3GPP decided it would stay that way.

And it’s growing. Omdia’s own tracker has NB-IoT commanding Asia and Oceania at 86 percent of global module shipments in 2025, and LTE-M expanding to a 58 percent regional share by 2035. NB-IoT, massive IoT and eRedCap are projected to make up around two thirds of cellular IoT connections by 2035.

So the industry is asking operators to name a death date for the radio carrying two of the technologies its own forecast says will still be growing nine years from now.

And there’s no cost gun to their heads. This is the part that kills it. 3G died partly because 3G and LTE need separate core networks, and running two cores is a bill that arrives every month, so operators had a hard financial reason to consolidate. That specific pressure doesn’t exist here. Modern 5G architecture doesn’t force the same separation between NR and LTE, and 3GPP has been working on letting NB-IoT and LTE-M attach to the 5G core as well. The duplicate-core bill that helped force the 3G decision never lands on the desk.

None of which makes LTE free. Operators still carry spectrum, power, maintenance, radios and vendor support lifecycles, and every one of those argues for shrinking LTE over time. Refarming will happen. But those pressures are gradual and negotiable in a way core consolidation wasn’t. They produce a slow squeeze, not a date.

So the deadline that would unlock eRedCap volumes isn’t late. It may be structurally unavailable.

Now look at what every transition in the section above had in common. Openreach is retiring the PSTN because the PSTN costs it money: an ageing network, faults climbing, hours lost. Nobody made it do that. Analogue television went because the spectrum was worth more empty than full. 3G went because the second core was a bill. In every case, the owner of the network wanted the network dead, and the date was the instrument.

LTE isn’t like that. LTE earns.

The 2G and 3G sunsets worked because operators wanted them. Nobody wants this one.

The spec is clear. The language isn’t.

A small thing, but it kept nagging.

Omdia’s July blog says eRedCap targets different use cases from RedCap. Coverage of the accompanying Cellular IoT Market Tracker says eRedCap inherits the use cases of its predecessor. Same firm, same season, opposite claims.

So I went looking for the disagreement in the standards, and it isn’t there. Telit, Ericsson and GSA all describe eRedCap as a distinct device type, capped to 10 Mbps, rather than a generational successor to RedCap. On the substance the vendors are lined up. The specification is not confused.

The confusion is entirely in the language wrapped around it. Trade coverage keeps reaching for “evolution”, “next step”, “successor”, because that is how our industry narrates a version number going up. Two years after Release 18 froze, a buyer can read three write-ups of the same standard and come away with three different mental models of what eRedCap is for.

That’s not a spec problem. It’s a vocabulary problem, and it costs more than it looks like it should, because a buyer who can’t confidently place a product doesn’t buy it.

He buys the full 5G module.

The back catalogue

Worth remembering who’s delivering this verdict.

In October 2024, RCR Wireless reported Omdia forecasting almost a billion RedCap and eRedCap connections by 2030: 963.5 million, at a 66 percent compound growth rate. The same piece noted Omdia had been bullish on RedCap for years, to the point of calling it the big missing piece of the 5G IoT puzzle.

By July 2025 that had softened to surpassing 700 million by 2030. The two figures are framed differently and I won’t pretend it’s a clean like-for-like. But “almost a billion” and “more than 700 million” are not the same sentence, and nine months is not long.

In December 2025, Omdia called the Apple Watch the starting gun for RedCap adoption.

By June 2026, eRedCap was expected to outperform RedCap.

This month, RedCap will struggle.

That isn’t a scandal. Analysts update when the evidence moves, and Omdia has been more candid about doing it than most. But read the current position as a correction rather than a prophecy. Anyone who built a roadmap on the 2024 version is entitled to feel slightly seasick.

The starting gun was a wristwatch

The Apple Watch detail is the tell, and the current research is admirably blunt about how it went: the watch gave RedCap some momentum over the past year, and the technology still never reached anticipated volumes.

Think about what that means. RedCap was designed for mid-tier industrial IoT. Sensors, cameras, trackers. The stuff between an NB-IoT meter and a full 5G gateway. Wearables were on the list but they weren’t the pitch. The thing that finally validated it was a consumer smartwatch, and even that wasn’t enough to hit the numbers.

Apple can absorb a module cost that a company deploying four thousand telemetry units cannot. Apple also has a two-year replacement cycle rather than a ten-year one, so the longevity anxiety that stops industrial buyers doesn’t touch them. When your proof point is the highest-margin consumer device on earth and the volumes still disappoint, you haven’t proved the mid-tier industrial case. You’ve demonstrated it never turned up.

RedCap didn’t fail. It turned up early.

RedCap arrived in 2023 into a world with very little 5G SA. No SA, no RedCap. The addressable market at launch was a fraction of the market it was designed for. If you want the Release 17 side of that story in detail, we cover it on 5gredcap.co.uk, along with UK operator deployment status and hardware.

eRedCap isn’t smarter than RedCap. It’s later. It arrives into networks that already exist, on simpler silicon, and it lands nearer the Cat-1 price point, where the future-proofing argument gets harder to make because the gap you’d be paying to close is smaller.

Same industry, same standard family, better timing. Nothing was wrong with the technology. Anyone reading this as a caution about reduced-capability 5G in general has taken the wrong lesson entirely.

What the summary leaves out

Two things, both of which will decide more deployments than module price.

5G SA being live is not the same as RedCap being enabled. This gets glossed constantly, and I’d gently point out that “5G SA networks now widely deployed” is doing some heavy lifting in a country where most 5G service is still Non-Standalone. A RedCap or eRedCap device needs the operator to have switched RedCap support on in the radio network. The base station advertises support in SIB1, identifies RedCap devices during random access via a dedicated initial bandwidth part and specific MAC control element logical channel IDs, and configures accordingly. It’s a software matter rather than new hardware, which is why it gets waved away, but a 5G SA network with RedCap features off will not serve your devices. 5G-ACIA states it plainly enough: adoption needs SA operation plus software updates. “SA is live there” is not an answer to the question you’re asking.

Roaming is nascent. GSA’s own material says RedCap roaming needs dedicated inter-PLMN agreements and RedCap RAT types introduced into core policy frameworks. If you’re deploying trackers across borders, that’s your constraint. Not throughput, not module price, not silicon. Anyone who has sold connectivity for a couple of decades recognises this instantly, because it’s the same problem in a new hat, and it’s always the one that surfaces three months after the pilot went well.

The standards calendar

Put Omdia’s curve next to 3GPP’s.

eRedCap adoption accelerates in 2028 and 2029. Release 19 froze in December 2025 and added no third reduced-capability device type. Release 20 is the final 5G Advanced release and already runs alongside the first 6G studies. Release 21 is expected to carry the first 6G specifications.

So the reduced-capability NR family looks complete at two, and the market will reach the second one at roughly the moment 3GPP finishes with the generation.

Which is, awkwardly, an argument for the people doing the overspecifying. If you’re the buyer who picks full 5G because you’re nervous about being stranded on a tier that gets superseded, the counter is that nothing is queued to supersede eRedCap. Ambient IoT gets described as the next rung down, but it’s a different radio paradigm rather than a further-reduced NR device, and its specifications don’t build on RedCap at all.

Two device types. That’s the family. Nobody is coming to obsolete you.

What I’d actually do

The rational play, on Omdia’s own analysis, is to wait. Modules launched early 2026, first devices expected 2027, prices falling into 2028 and 2029. With no announced switch-off, there’s no penalty for waiting.

The risk is that everybody runs the same calculation, and that the trigger they’re all waiting for never gets pulled. If the LTE date is structurally unavailable rather than merely late, then “wait for prices to fall” is waiting on volumes that are waiting on urgency that is waiting on a deadline nobody has a reason to set. That’s not a queue. That’s a standing wave.

If you have a decision this year, the questions in order are: does my target operator have SA live and RedCap enabled, does this deployment cross borders, and what’s my device lifetime. Single-market, fixed, long-life, RedCap enabled on your operator: engineer for eRedCap now even if you ship Cat-1bis in the interim. Crossing borders: the roaming answer lands before the price answer does, and waiting for 2028 doesn’t fix it.

And if the Cyber Resilience Act is in scope for your product, note that you already have a dated obligation pulling you toward eUICC. That deadline is real in a way the 4G sunset currently isn’t. Build the SIM architecture around the deadline that exists, not the one that doesn’t.

Which means eRedCap’s case has to stand on its own economics rather than on a sunset. Cheaper silicon, simpler design, lower power, 5G core features. If those are worth it to you at 2027 prices, buy it.

If your business case needs the 4G sunset to close, your business case has a dependency it cannot schedule.

And if anyone shows you a forecast for 2030, ask them what they were forecasting in 2024.

---

Sources