Has Cellular IoT Failed? Looking Beyond the Headlines After AT&T’s NB-IoT Network Shutdown

While operator network shutdowns are not unusual, at first glance, AT&T’s choice to shut down Narrowband Internet of Things (NB-IoT) might seem like an arbiter of something bigger. But let’s put this into perspective. Operators have to make decisions all the time on which cellular technologies to invest in. Historically, that has been driven by their handset business, so the Internet of Things (IoT) has piggybacked off handset network choices. NB-IoT was the first network that was chosen primarily for IoT use. When NB-IoT devices first became available around 2017, there was much hype about NB-IoT for the massive number of connections it could enable. China’s first mover choice with this technology and its own over-forecast use of it caused the world to take notice and consider investment.

What made it attractive to operators is that the investment in NB-IoT was minimal. NB-IoT was a software layer (for most network infrastructure equipment) and could leverage spectrum available to operators—either in Long Term Evolution (LTE) guard bands, in available LTE spectrum blocks, or using 2G spectrum after 2G networks were shut down.

With nearly 8 years of practical use of NB-IoT, its main challenge has been that it did not become a 2G network replacement technology. First, unlike 2G, NB-IoT does not offer global coverage today, partly because Cat-M is also a strong candidate for low data use cases and many operators worldwide have chosen to only deploy Cat-M. As a result, NB-IoT offers spotty geographic coverage, which is very evident in Europe. Second, roaming was never optimized for three reasons. One reason was because network parameters for the technology were not consistent from operator to operator. These parameters included network check-in times and power-saving mode durations. The second reason was that carrier billing platforms were not optimized for NB-IoT message-heavy data collection where the end result is a small volume of data exchanged. With roaming fees based on data volume (not messages exchanged), roaming profitability was limited, so why bother building NB-IoT roaming agreements. Third, NB-IoT was failing in asset tracking use cases, a potentially high-growth application, as the technology cannot be engineered (no matter how hard an operator tries) to meet mobility requirements (tower hand-offs) and keep battery life long.

The bigger question concerns cellular’s place in the IoT value proposition. Not surprisingly, cellular cannot address every use case but cellular’s share of the IoT pie is effectively flat. As shown in ABI Research’s Internet of Everything Market Tracker (MD-IOE-113), cellular held a 4.4% share of the IoT installed base in 2020; by 2028, this share is 4.3% (albeit on a bigger base). One reason is that there is a growing number of use cases that can be served by local area connectivity. Short-Range Wireless (SRW) technologies such as Bluetooth® and Wi-Fi are evolving with more capabilities across range, bandwidth, and power consumption. In MD-IOE-113, Bluetooth® and Wi-Fi’s share of the IoT installed base in 2020 was 30%; by 2028, that is expected to be 36%. (Note that cellular, Bluetooth®, and Wi-Fi have all grown their installed base of connections by over 3X between 2020 and 2028, and in 2024, each has billions of connections.)

The second reason is the cost of cellular. You have to pay for access to a cellular network, in addition to the device itself. With Bluetooth®, Wi-Fi, and others, spectrum access is free, and building a network is not expensive (relatively), particularly when considering that SRW network value also comes from supporting connectivity to smartphones and tablets. The bottom line is that cellular Wide Area Network (WAN) technology is not needed if the necessary network coverage is a Local Area Network (LAN) or even a Neighborhood Area Network (NAN). In addition, there are other WAN technologies that can support LAN and NAN connections to the cloud; and hundreds of LAN and NAN connections can be supported by a single WAN connection.

The third reason is that unlicensed spectrum is suitable for many IoT use cases. As the network coverage area shrinks, interference can be better controlled. Many IoT use cases also don’t require low-latency and high-reliability message delivery.

Finally, new technologies such as New Radio+ (NR+) (DECT 2020 NR) based on mesh networking can better utilize unlicensed spectrum and achieve high reliability and low latency, even in larger NAN areas. As a result, some of the benefits of cellular using licensed spectrum are diminished.

While NB-IoT has been beaten up as of late, two developments will improve its stature among the connectivity elite and help drive growth. First is the Non-Terrestrial Network (NTN) specifications in The 3rd Generation Partnership Project (3GPP) Release 17, enabling Narrowband (NB) protocol message delivery over a satellite link. With this development, the IoT Total Addressable Market (TAM) will be expanded to the sparse or uncovered areas of cellular to serve markets such as oil & gas infrastructure, agriculture, and smart grid. A key factor driving NTN growth is that many smartphones already support some form of this technology. For example, Google’s Pixel 9 follows the NTN standard with NB-IoT, while Apple iPhones use a proprietary version. Additionally, SpaceX is about to launch its own version, enabling text messaging outside of cellular coverage for any device hosting a partner’s Embedded Subscriber Identity Module (eSIM). The IoT has always benefited when the smartphone and tablet supplier ecosystem is also using the same connectivity technologies, driving more scale to lower prices and enabling innovation. (Note: NTN is a recent technology, and its full impact has not been reflected in MD-IOE-113.)

The second development is that NB-IoT is gaining traction in the water metering market. This may only be a European phenomenon, but this market’s growth is due to operators assuring the meter Original Equipment Manufacturer (OEM) community that they will support this technology for the next 15+ years. And this is a commitment because existing 4G NB-IoT devices will not easily transition onto 5G Standalone (SA) networks, even if NB-IoT has been included in the 5G 3GPP specifications. Regardless, the key point is that for any application segment, the ecosystem needs to demonstrate commitment to technology if it expects customers to invest.

NB-IoT and, more broadly, cellular are tools in the toolkit for IoT applications in a market that is increasingly accepting multi-technology devices and network architectures to best serve the IoT use case. NB-IoT best serves markets that are low data and require wide area coverage. It does better when devices are stationary, but its low device cost makes it suitable for asset tracking when tracking frequency is low. In fact, fueling NB-IoT growth will be smart labels designed to be very cheap without rigorous location requirements. ABI Research expects NB-IoT to be a leading connection technology for cellular IoT, shipping 110 million devices in 2024, increasing to 190 million in 2030. NB-IoT has not become the nirvana that Low-Power Wide Area (LPWA) technology was hyped to be, but NB-IoT and cellular overall are far from dead!