Latest / The 5G Podcast with Fexingo: Wireless Networks, Carriers, and Mobile Infrastructure / Why Cellular IoT Is About to Overtake Wi-Fi for Low-Power Devices
Transcript
- Lucas: So here's a number that stopped me this week: by the end of 2025, cellular IoT connections — think smart meters, asset trackers, industrial sensors — hit roughly four point three billion globally. That's up about 22 percent year-over-year, and the really interesting part is that for the first time, the average cost of a cellular IoT module dipped below the average cost of a Wi-Fi module for low-bandwidth applications. Luna: Wait — cellular is cheaper than Wi-Fi? For years the pitch has been that Wi-Fi is basically free once the router's there. What changed? Lucas: What changed is that the chipset vendors — Qualcomm, Sony Altair, Sequans — they've been driving the bill of materials down for lte m and nb iot modules. You can now buy an nb iot module in volume for around one dollar fifty per unit. A comparable Wi-Fi module with similar power draw is still around two bucks. And that's before you factor in the infrastructure costs. Luna: Right, because Wi-Fi needs a gateway, a power source nearby, a network setup. Cellular just needs a SIM and a signal. Lucas: Exactly. And that's the massive IoT thesis — you don't want to run Ethernet or maintain a Wi-Fi mesh for ten thousand sensors spread across a farm or a factory floor. You want them to just work on the existing cellular tower. The battery life on nb iot can hit ten years with a couple of AA cells. Wi-Fi drains faster because the protocol wasn't designed for that kind of sleep cycle. Luna: So we're seeing a real crossover. I read that T-Mobile added over eight million IoT connections in Q1 this year alone, most of them on lte m. They're clearly betting this is a growth vector beyond phones. Lucas: They are. And it's not just T-Mobile — Verizon and AT&T are both pushing their own IoT platforms. But the interesting angle is that this isn't about 5G's high-speed promise. It's about the low-power, wide-area part of the standard — what 3GPP calls 'massive machine-type communications.' lte m and nb iot were actually defined in 4G-era releases, 12 and 13, but carriers are now deploying them at scale because the economics finally work. Luna: And 5G brings a new variant — nr light, or reduced capability. I think that's targeted at wearables and mid-tier IoT that needs more bandwidth than a temperature sensor but less than a video stream. Lucas: Yeah, nr light is fascinating. It's essentially a 5G modem that's been stripped down — lower antenna count, narrower bandwidth — so it can hit around 150 megabits per second downlink. That's enough for a smartwatch to stream music or a security camera to send 1080p video, but the module cost should be around five to eight dollars, compared to twenty-plus for a full 5G modem. That opens up a whole middle tier of IoT that's too heavy for nb iot but too light for premium 5G. Luna: Let's ground this in a real example. You mentioned German utility earlier — what did they do? Lucas: So there's a municipal utility in Bavaria, Stadtwerke something — they manage about fifty thousand water meters across a region with a lot of rural coverage gaps. They originally used Wi-Fi mesh with battery-backed relays. The problem: batteries in the mesh nodes died every eighteen months, and the relay units were expensive to maintain. They migrated to nb iot on Deutsche Telekom's network last year. Each meter now reports consumption once a day with a one-second transmission burst. Battery life target is twelve years. They've already cut operational costs per meter by about 70 percent because they don't send a truck out to change batteries every year and a half. Luna: Seventy percent is huge. And that's without even talking about the data value — consumption patterns, leak detection, dynamic pricing. Lucas: Right. The meter data itself becomes a revenue stream. But the infrastructure story is the one I find more interesting, because it's a template for other verticals. Agriculture — soil sensors, livestock trackers. Logistics — pallet and container tracking. Smart buildings — HVAC sensors, air quality monitors. All of these have similar profiles: low data, long battery, spread-out geography. Luna: There's a tension though. Carriers are used to selling high arpu smartphone plans. A water meter that sends a few kilobytes a day might generate — what — fifty cents a year in revenue? That doesn't exactly light up the income statement. Lucas: You're right. The per-connection revenue is tiny. But the volume is enormous. Ericsson's mobility report from November projected around seven billion cellular IoT connections by 2030. If each one generates, say, two dollars in annual service revenue — that's fourteen billion dollars in incremental top line. And the marginal cost for the carrier is almost zero once the network is built. The spectrum is already there, the towers are already up. It's basically pure margin on those additional connections. Luna: And we should mention that this doesn't require 5G standalone — lte m works on existing 4G infrastructure. So carriers can capture this revenue without a massive 5G upgrade spend. Lucas: That's key. lte m and nb iot can coexist on the same spectrum block as 4G voice and data. Carriers are basically monetising a radio resource that was already deployed. And for the first time, the module cost advantage has flipped in favour of cellular over Wi-Fi for low-power use cases. That's the structural shift we're tracking. Luna: If today's episode was useful to you and you want to keep it ad-free, buy me a coffee dot com slash fexingo helps. Lucas: Yeah, that really does make a difference. We don't run ads — that's a deliberate choice. If you want to support that, buy me a coffee dot com slash fexingo is the place. Luna: Alright — back to the shift. So if cellular IoT is now cost-competitive at the module level, what's the bottleneck? Why isn't every utility and farm switching overnight? Lucas: The bottleneck is the ecosystem maturity on the application side. The hardware is cheap, the network is ready, but the backend — the device management platform, the data pipeline, the integration with existing enterprise systems — that's still clunky. Most utilities run on decades-old SAP installations. Connecting an nb iot stream to that back-end isn't trivial. Companies like Sierra Wireless and Telit are building middleware to bridge that gap, but it's still early. Luna: And there's a security concern too. Having millions of always-connected devices on a carrier network — that's a huge attack surface. The same carrier that never patched your router might be the one managing your water meters. Lucas: That's a real worry. Carriers are not historically great at IoT security. The SIM authentication is strong — it's based on the same standards as banking SIMs — but the device firmware and the cloud backend are often weak links. We've already seen botnets built from IoT devices. As the installed base grows, that risk scales. The GSMA has a set of security guidelines for IoT, but compliance is voluntary. Luna: So the opportunity is real, but the execution requires carriers to think differently. They have to become platform companies, not just pipe providers. Lucas: Exactly. The winners in massive IoT will be the carriers that offer a full stack — connectivity plus device management plus data analytics. T-Mobile's 'IoT for Good' program is a small example; they give free connectivity to nonprofit sensors monitoring air quality. That's a branding play, but it's also a way to build the ecosystem. Over time, those relationships turn into commercial contracts. Luna: And nr light will add another layer. We'll probably see smart glasses, advanced wearables, maybe even some automotive telematics use it. That's the next wave. Lucas: Right. The reduced-capability 5G modems will start sampling in volume later this year, with commercial products likely in 2027. So we're probably two or three years from seeing nr light in consumer devices. But the infrastructure for massive IoT is already here. The crossover we talked about — cellular cheaper than Wi-Fi — that's happening right now in procurement contracts. Luna: One more angle: the chipset makers. Qualcomm has the lte m modules, but MediaTek and especially Chinese firms like UNISOC are pushing prices down. Could we see a race to the bottom similar to what happened with Wi-Fi chips? Lucas: That's likely. The module price has already dropped from about five dollars in 2020 to roughly one fifty today for nb iot. In a commodity market, margins compress. The value moves up the stack — to the platform, the data analytics, the vertical solution. The chipset becomes a pass-through cost. That's why we see companies like Semtech acquiring IoT platform companies — they want to own the relationship with the enterprise, not just sell a radio. Luna: So the long-term winner might not be the carrier or the chip vendor — it could be the middleware player that makes it simple to connect a sensor to a dashboard. Lucas: That's my bet. The physical connectivity is becoming a utility — like electricity. The value is in the applications that run on top. And right now, that layer is fragmented. There's an opportunity for a company to become the AWS of IoT connectivity — a single API that works across carriers and module types. Nobody has done that yet at scale. Luna: We'll be watching. Thanks Lucas — this one gave me a whole new way to think about what 5G actually means beyond faster downloads. Lucas: Happy to share it. Next episode we'll look at how edge computing is changing the latency story for industrial IoT — that's another angle that ties directly into this massive IoT conversation. Luna: Looking forward to it. For The 5G Podcast, I'm Luna. Lucas: And I'm Lucas. Catch you next time.