Cellular IoT Trends from the Show Floor at MWC26 Barcelona and embedded world 2026

The Internet of things (IoT) has transitioned from being a topic that was publicly front-and-center in carrier connectivity strategies at Mobile World Congress (MWC) to being an implicit underlying enabler instead. The IoT was always something that was selectively referenced by carriers, conveniently excluded whenever it diluted Key Performance Indicators (KPIs), but then also promoted whenever it could be used as a sign of new growth, and/or market leadership. Most carrier stands at MWC26 Barcelona made little to no direct reference to the IoT, and examples of connected autonomous implementations were limited. Interestingly, any such demonstrations tended to be themed around robotics and drones; employed in various industrial and environmental applications where mission-critical wireless connectivity was simultaneously vital, while also being one part of a greater whole that included edge compute, and Artificial Intelligence (AI).

Edge AI was also the most important topic at embedded world the week after, where there was effectively no discussion of AI data center products and cloud-based intelligence. For the sake of security, enterprise ownership, and control, all practical demonstrations of AI were shown in an edge deployment context. Remote services were presented as a powerful means of model training, but with model execution being local. This did not eliminate the importance of Wireless Wide Area Network (WWAN) connectivity; however, the principal post-training remote services presented were Firmware Over-the-Air (FOTA) updates and security patches. Edge Artificial Intelligence of Things (AIoT) does not diminish the importance of connectivity. It may, in fact, enable both local and wide area device connections that could not be deployed before, as a purely cloud-based AIoT implementation may have data transport costs that rendered a system uneconomical.

Narrowband-Non-Terrestrial Network (NB-NTN) and New Radio (NR)-NTN remained of keen interest on the part of carriers and semiconductor manufacturers at MWC26. NTN is an opportunity to fill gaps in technical capability and physical coverage. For the IoT specifically, NTN exists to sell connectivity and equipment where it was not possible before. For semiconductor manufacturers, NTN can be enabled by a firmware upgrade to many existing cellular chipset designs—a minimal investment. For carriers, satellite connectivity is treated the same way as any other roaming agreement. The initial logistics may be more complicated, but if customers do not buy into NTN, terrestrial carriers lose nothing, as they will consequently have nothing to pay their satellite partners when roaming reconciliation occurs. The biggest cost falls on the device Original Equipment Manufacturers (OEMs) that must integrate NTN chips, and associated filters and antennas required, for a market that is currently unproven. It makes some sense to argue that all cellular modems should include NTN, and that all cellular carriers should have satellite partners, but long-term commitment remains uncertain when the stakes seem so low.

Cat-1bis continues to dominate as the cellular connectivity technology of choice in the IoT, even in new partnerships and product launches. This is despite the limited window of opportunity for Cat-1bis, as Long Term Evolution (LTE) sunsets loom. Nordic Semiconductor, which built its business, in part, on the integrity of a 100% non-Chinese supply chain, has partnered with China’s Eigencomm to launch its nRF93M1 module. This is in anticipation of Nordic entering the enhanced RedCap (eRedCap) market, which will replace Cat-1bis. Nordic’s reputation is so strong that OEMs and carriers trust it to sell a secure product during what remains of the Cat-1bis window. Other Western vendors, such as Telit, will continue to invest in Cat-1bis, especially in single-mode products, to be as low cost as possible, and sell as much volume as possible in the build-up to eRedCap. With the exception of Qualcomm, all IoT semiconductor manufacturers have announced their eRedCap intentions. Only MediaTek has deliberately chosen not to develop a product.

Second-generation 5G RedCap chipsets are not coming soon, according to Qualcomm. Not until the 5G RedCap market takes off, proving that further investment is worthwhile, and certainly not before eRedCap chips are launched. The most frequent criticism of 5G RedCap has been that it is too expensive at every downstream point because the chips are too expensive, some models being cut-down versions of enhanced Mobile Broadband (eMBB) designs. But it seems there is actual progress in 5G RedCap adoption, and that it won’t end up as a mere technology project, even if just in select markets for now, namely the United States and China. Module vendors are seeing demand in North America for video surveillance, and mid-tier gateways for failover, fiber replacement, and public safety. There are also some discussions in Europe about using 5G RedCap in smart meter gateways. While in China, it has been suggested that up to half a million RedCap chips may have been sold into the IoT in 2025, for smart meter gateways and Telematics Control Units (TCUs) (aka “T-Boxes”). Millions more 5G RedCap chips have purportedly been sold in China in 2025 for smartwatches, Customer Premises Equipment (CPE), and MiFi.

SGP.32 was everywhere at MWC26, not just with regard to smart card and digital security vendors, but every party connected to the cellular IoT value chain. The implication is that the increased ease of assigning Subscriber Identity Module (SIM) credentials stands to boost the market for cellular connectivity by simplification. But embedded SIM (eSIM) may not necessarily be more cost-effective than traditional SIM cards, outside of scenarios where the SIM needs to be replaced with that of a new service provider. This places eSIM in the position of eliminating indecision more so than eliminating a lack of technical capability or affordability. It is unclear if eSIM increases the addressable market for the IoT, i.e., enabling connectivity where it was not possible before; or if eSIM is a just migratory technology, not augmenting cellular market potential, but gradually assuming control of it. eSIM seems to be perpetually plagued by the shadow of a better version in the offing in the near future—in the form of SGP.42. Meanwhile, at embedded world, there was a more specific focus on the IoT on the European Union’s (EU) Cyber Resilience Act (CRA), and what device OEMs need to do to remain compliant.

The Random Access Memory (RAM) component shortage was an omnipresent topic at embedded world. The vast amount of RAM being supplied to data centers risks starving IoT device OEMs. Numerous IoT component vendors have stated they were forewarned by COVID-19, and have built up an inventory of RAM and sourced secondary suppliers. This caused an increase in their costs, but no shortages. Many IoT components do not require large amounts of RAM per unit. But the situation is different for their downstream device OEM customers. Without discrete supplies of RAM to complete their own devices, OEMs will have no product to sell and no need/ability to order more from their other upstream component suppliers. Bigger device OEMs are prioritized by RAM manufacturers, but some are convinced that consolidation must happen, and that some may go out of business. The over-invested-in AI data center business is strangling the edge IoT device business, which is the real, mission-critical market.