Hello readers!

This week, we’re talking about how satellite connectivity is moving to the chip, why the smartest buildings won’t have the most devices in them, and more!

For most of IoT’s history, a device that needed to work beyond the cell tower needed a second identity. It carried a second radio, often a second SIM, a second certification path, and a second line on the bill of materials, and engineers justified all of it one deployment at a time. That arrangement is ending. Satellite connectivity is becoming something a standard cellular IoT chipset can handle on its own, which moves the coverage question out of hardware design and into the connectivity plan: which networks a device is cleared to use, under which contracts, and how it behaves when it switches between them. The cost of reaching a remote asset used to be paid up front in components. More and more, it will be paid in decisions made after the device ships, and teams that still treat satellite as a separate product category are likely to scope those decisions wrong.

The old arrangement had a logic to it. IoT Analytics has noted that satellite has carried asset tracking and telemetry since the late 1970s yet was long treated as a last resort behind terrestrial networks, held back by cost and power efficiency. Hybrid satellite-terrestrial products did exist, but they needed two separate RF chipsets in the device or terminal. Coverage kept the question alive anyway, since the same analysis put terrestrial networks at roughly 15–20% of the planet’s surface. For a smart meter bolted to a house, that number never comes up. For a shipping container, a pipeline sensor, or a herd of cattle, it decides whether the deployment works at all. IoT Analytics also pointed to an alternative: a single RF chipset handling both terrestrial and satellite links, possibly added to existing devices through a firmware update with little or no hardware change. That alternative is now reaching mainstream cellular silicon.

This month, Iridium announced that Altair Semiconductor is building Iridium’s NTN Direct service into its ALT1350 and next-generation ALT1550 chipsets, letting device makers reach Iridium’s LEO constellation without adding a separate satellite modem. The ALT1350 is a low-power LTE-M/NB-IoT part aimed at asset tracking, smart metering, agriculture, and vehicle telematics. The ALT1550 extends the same option to 5G eRedCap devices such as wearables. Iridium’s service runs on 3GPP standards and globally coordinated L-band spectrum, and Altair joins a growing roster of chipmakers, module makers, and OEMs supporting it. Altair has made this argument before, noting in 2024 that cellular networks reach more than 80% of the population but less than 40% of land and that falling costs were pulling non-terrestrial networks out of their last-resort role. Two years later, it is shipping satellite as one more mode of a mainstream cellular chip.

Folding satellite into the chip removes a hardware problem and exposes a sorting problem. “NTN” covers several technologies that don’t interoperate, and Monogoto argues that spectrum, more than orbit, decides which networks a module can use. 3GPP Release 17 defined two IoT-NTN bands, n255 in L-band and n256 in S-band, and a device certified for one won’t attach to a constellation running the other. Orbit adds a second axis. The NB-IoT NTN service that is commercially live today runs mostly over GEO through Skylo, with LEO services from Iridium, OQ Technology, and Sateliot now coming up alongside it. Direct-to-device offerings from Starlink and AST SpaceMobile borrow terrestrial operator spectrum and are built mainly for phones, not sensors. Monogoto also separates radio compatibility from commercial access, and that distinction carries more weight once the radio comes built in. A chip that can talk to Iridium still needs an agreement that lets it. Choosing which of those agreements a device carries, and when it’s allowed to use them, now falls to the team that buys the chip.

Ask an IoT team how to make a physical space smarter and the answer usually starts with a bill of materials. Devices to mount. Power to run. A network for them to join. A maintenance schedule someone has to own. That instinct built this industry, and for a great many problems it is still the correct answer.

Read more

In this episode of the IoT For All Podcast, Patrick Ferriter, SVP of Product at Agora, joins Ryan Chacon to discuss building real-time voice AI experiences, even on bad networks. The conversation covers what physical AI means, moving from controlled demos to the real world, architecting a voice AI experience for high latency networks, physical AI uses cases, AI companions, using edge or cloud for AI, privacy and security, and the future of conversational AI devices.

YouTube • Spotify • Apple • YT Music • Amazon