LoRaWAN has reached 125 million deployments, growing at 25 million per year. Which still looks modest next to cellular IoT’s mega count. But LoRa Alliance chief Alper Yegin argues the comparison misses the point: LoRaWAN’s growth is driven by fast-moving private networks, better industrial integrations, and a starring role in the ‘physical AI’ economy. Note, this article is continued from here.
In sum – what to know:
Cellular comparison – While operators like Vodafone count hundreds of millions of IoT connections across cars, payment terminals, security systems, and industrial assets, LoRaWAN is carving out a complementary role in low-power, low-cost, and hard-to-reach applications. Major operators including Swisscom, Orange, Verizon, and AT&T are using both technologies side by side.
Private networks – The alliance says enterprise deployments are expanding faster than public operator networks, fuelled by specialist solution providers in sectors ranging from agriculture to school safety. A new three-year roadmap aims to accelerate adoption through industrial integrations, plug-and-play deployment improvements, and expanded satellite and roaming capabilities.
Tech symbiosis – AI could be the catalyst IoT has been waiting for. Yegin argues that so-called “physical AI” is fundamentally an IoT story: connecting sensors, devices, and machines to AI systems that can understand and act on the physical world. Combined with LoRaWAN’s open ecosystem, low-cost infrastructure, and broad application reach, he believes AI will help drive the next phase of growth for the technology and the wider IoT market.
So where were we? Well, we were trying to put claims by the LoRa Alliance – about the size (125 million deployments) and growth (25 million per year) of the installed base of LoRaWAN-based IoT devices around the world – into some kind of context. We are not quite done. Vodafone, arguably the leading cellular IoT provider, claims to have 240 million IoT connections on its books. So one tech via one operator – versus the whole LoRaWAN world.
Alper, just for the record: make sense of that mismatch for us – 125 million, total, versus 240 million by Vodafone alone. “Yeah. So that figure includes connected cars, ATM machines, point of sale devices, home security backup lines, software as well.” This is Alper Yegin, chief executive at the LoRa Alliance, speaking a couple of weeks back, on the release of the alliance’s technical roadmap for LoRaWAN development over the next three years.
The question is a little unfair; it’s not apples for apples – on the grounds Vodafone is pitching a family of cellular based IoT technologies, going from pulse-like NB-IoT for meters to full-fat 5G for private networks to serve elements of Industry 4.0. But he continues: “That’s how it has achieved that volume. It’s a more mature market. For Vodafone, these are also more profitable markets (verticals) – with higher volumes and ARPUs; and that’s a fair game.
“I would’ve done the same. You’re not going to use LoRaWAN to connect a car; you need higher-bandwidth LTE-M for that, at least. But there is a long list of IoT applications. Some operators are going after them and some aren’t. Swisscom is connecting bikes – like we said. Verizon and AT&T are using LoRaWAN, as well – alongside LTE-M. It is pick-and-choose, case by case. AT&T launched its Connected Spaces service a month after it terminated NB-IoT.
“Verizon has built a LoRaWAN network at the NBA Arena in Atlanta for the same. So whether they have a nationwide LoRaWAN network like Swisscom, KPN, Orange, Digita Oy, SK Telecom, Hutchinson Three in Austria, or they offer private LoRaWAN networks via their business arms, or they offer both – many see the complementary nature of LoRaWAN. And some aren’t touching it, of course.” The point is not to compare – except with the also-rans.
Versus the rest
And LoRaWAN has critical mass and momentum versus all of them – including the likes of MIOTY, Sigfox, Wirepas, Wi-SUN, Zigbee, Z-Wave, and so on, says Yegin – just by virtue of a well-oiled developer ecosystem, reaching parts of the IoT market that cellular IoT misses, in more varied fashion than any rival solution using unlicensed bands. The fact that traditional mobile operators see its value, variously, is telling. But they are just one sales channel.
Yegin explains: “Since the start, the strategy has been to deploy in four waves, in four set-ups: as public networks, like we see with the mobile operators; then in private networks, in enterprises; then in community networks, like with TTN initially, and Helium; and then in satellite networks, with Lacuna, Plan-S, Kinéis, now Fossa Systems, and so on. These waves are all working now.” Where is the most activity – which is driving that 25 percent growth hardest?
“Right now, the longest runway and fastest momentum is the second: the private networks. Which has two things going for it, right now: the breadth of the application space, and the agility (of the developer ecosystem).” He references Halter again (see previous instalment), the New Zealand agri-tech start-up that has just raised a $2 billion investment, including from Peter Thiel’s Founders Fund, as well as Atlanta-based panic button vendor CENTEGIX.
“Halter has a real sense of urgency; it is just sprinting ahead. Similarly, CENTEGIX has a million panic buttons, mostly in North America, mostly in schools. There are tons of these examples, with volumes going upwards. That is where we have the biggest numbers right now. While operators like Swissom are densifying their networks, and adding applications, we have this organic and extremely bullish ecosystem-force which is charging forward.”
Just quickly: what about Helium, the innovative “people-powered” community IoT proposition that has just been snaffled up (in some form) by Noble Mobile in the US? The deal is fresh, at the time of talking, but Yegin’s read on it appears to be right: that the deal is for Helium Mobile, the consumer MVNO business, with mobile subscribers, and not for Helium Network, responsible for its original LoRaWAN infrastructure, plus Nova Labs and HNT ecosystem.
Noble Mobile has also committed to continue using the Helium Network, rather than replacing it. Yegin says: “It sold the cellular operator business, and maintained the rest.”
Keeping pace
What matters now, more than historical numbers alone, is whether LoRaWAN can keep pace. Which brings us back to the alliance’s new technical roadmap, a three-year development plan to sharpen its position as a “fourth pillar” for IoT. The headline items are practical rather than revolutionary: tighter integration with established industrial and utility systems, simpler deployment and onboarding, broader coverage options via satellites and new spectrum bands.
It also contains a clutch of security, certification, and management upgrades. Yegin bundles the work in three tracks. The first task is to integrate LoRaWAN with older industrial application domains. The flagship work is a collaboration with the OPC Foundation to define how OPC UA, a “modern version of Modbus”, runs over LoRaWAN. Integrators have been manually integrating them; the OPC Foundation approached the alliance to standardize the process.
The two have a working group to define the specification, to be released shortly. “It is a win-win; they solve the integration, and we grab another piece of land,” says Yegin. “We did the same with electric metering and water metering, with DLMS and OMS over LoRaWAN. We did it with the IETF by adapting IPV6 to run over LoRaWAN, and open up new IP applications.” The alliance is also to integrate with the UI-1203 protocol for water meters in the US.
The second track is about making LoRaWAN easier to deploy and operate. The roadmap talks about zero-touch onboarding, device migration, capability discovery, and other plug-and-play improvements. Hare, ‘zero-touch’ is a target concept – about “continuous improvement until the setup is telepathic”, he laughs. The model is Apple’s AirTag: remove the tab, press the button, and the device appears on the network. Of course, IoT is not so simple.
The challenge is to bring something like it to industrial IoT, where installation and integration complexity still acts like a brake on adoption. “Building a standard is the first step; making standard-compliant products easy to configure is the next.” The third stream concerns coverage. Satellite integration features prominently, alongside support for additional frequency bands, and improvements to roaming between public, private, and community networks.
Another thing: AI
We are 40 minutes into the conversation, and we’ve not mentioned AI. Which is good, actually. But the IoT crowd has lots to say about, we know – with all the buzz about physical AI and the enterprise edge. Which, to an extent, is the same IoT script as five years ago. But AI is not a stated part of the roadmap – so what’s the take on it? He responds: “Yeah, the AI world is turning its attention to so-called physical AI, which is just a fancy way of talking about IoT.
“Because it just means pulling data out of sensors in the physical world, taking it to AI systems for processing somewhere, and sending commands back to the physical world, whether to close a valve or direct a vehicle. Which is IoT, literally – as the building block for the AI to interface with the physical world. And LoRaWAN, with the widest application set, is at the forefront of making it a reality.” Edge AI, the middle function, runs the IoT gamut, he says.
He references vibration and camera sensors from the likes of Advantech, Honeywell, Milesight, NKE Watteco, and TE Connectivity. “This is deep-edge, on the device itself.” There is some processing in the core network, too, rendered in dashboards to monitor traffic to optimize data in flux and enhance third-party apps. “It’s like bringing an additional brain,” he says. Besides, the community is using AI to make better sense of IoT data – for use by AI.
“You have to be like an F-35 pilot to read some of these IoT dashboards, right? But AI brings native interactions – with generative text and voice queries to interrogate alerts. So the other angle is that AI makes IoT easier to use, and easier to get value from – which will accelerate adoption of IoT as well.” Back to the start, about the community being more hopeful and confident than ever – it seems like its biggest advantage for LoRaWAN is just that: the ecosystem.
Is that right? “Well, the ecosystem is definitely a part of it. I mean, it’s not just down to the technology itself. There could be better technologies, which haven’t succeeded because they don’t have the same ecosystem. But the ecosystem has only grown up around LoRaWAN because the decision was made to make the technology open. I mean, Sigfox wanted to be the Apple of IoT, right, and call the shots. But IoT cannot work like that; it is such a beast.
Taming the beast
“You cannot tame it, you can’t box it in. You have to let it grow, and to help it to grow. So the openness is the thing; plus the accessibility – because LoRaWAN uses unlicensed bands, which puts it ahead of cellular. Because you can build a network at home. It’s unconstrained. You don’t need a license. With cellular, you can’t touch it – you and I cannot touch it. Plus, the infrastructure is low-cost – which is not just by magic. It is designed that way.
“The price of a LoRaWAN base station, indoor or outdoor, is pretty close to a Wi-Fi access point. Again, because we designed the technology and architecture that way – long range, so you light up your home, plus your whole neighborhood, and valuable to many applications, so it works for the rhino tracker, the ranger vehicle, the security sensor, the water meter, the weather station. One network, multiple applications. Which is different to the rest.”
So why is nobody making money at this? Doesn’t this democratization of technology, by cost and usage, make it somehow a little too worthy for get-rich entrepreneurs? “No, I wouldn’t say that. I mean, IoT has been around for four decades. AI has been around even longer, and is just now seeing crazy take-up. So technologies follow their own cycles, and IoT has been difficult – many pieces need to come together to make a whole solution work.
“But we are entering the zone now; the curve is steepening. I keep going back to the examples – like Halter and CENTAGIX, which show there are real commercial opportunities, which can be rapidly capitalized at a significant scale. And I’ll go back to your earlier question: what did we get wrong? I wouldn’t say we got anything wrong; I just think the industry underestimated how fast the market would absorb it. Which is about awareness and education.
Those are the speed bumps at the moment. It’s not the competition, it’s not really the technical capability. This roadmap is exactly in line with the market’s awareness and appetite. Which just takes time – but is also being accelerated by this AI story.”
So there, it is about AI, as well.
