IoT Cellular Modules Compared: u-blox, Quectel, Nordic, Sierra (2026 Guide)

Cellular IoT is the connectivity layer with the steepest hidden cost — the module is cheap, the certification is not, and the SIM management is a 10-year operational decision. Here is what to know in 2026.

What you are actually picking

A cellular IoT module bundles four things into one shielded brick:

1. The radio modem (LTE-M / NB-IoT, sometimes 4G fallback)
2. A baseband processor — typically a Cortex-M4 you can run application code on
3. A SIM interface — physical, eSIM (eUICC), or increasingly iSIM (integrated)
4. Carrier certifications — the regulatory and operator approvals that decide whether the module can legally connect on a given network

The module choice locks in coverage maps, certification paths, and software toolchain — for years. Boring decisions matter here.

The vendors that actually ship

u-blox SARA-R510 / R5 series

The default for European and global deployments. Mature AT command set, good documentation, broad operator certifications. Pricey but reliable. SARA-R510M8S adds GNSS in the same module — convenient for asset tracking.

Quectel BG95 / BG770A

The volume play. Lower BOM cost, broad certification coverage, the module most low-volume Asia-and-EU products ship on. Documentation is improving but lags u-blox. Quectel’s eSIM / iSIM offerings are increasingly competitive.

Nordic nRF9151

Nordic’s cellular module family. Combines a Cortex-M33 application processor with an LTE-M / NB-IoT modem in one package — no separate MCU needed for many products. Native Zephyr support, and the developer experience matches what Nordic ships on its BLE modules.

Sierra Wireless HL78xx

North American workhorse. Strong AT&T / Verizon / T-Mobile certifications, robust operating temperature range, common in industrial and fleet products in the US. Heavier module footprint than u-blox.

Murata Type-1SC

Compact, low power, often picked for wearables that need cellular fallback or for products where space is the binding constraint.

How to pick

Walk through these questions in order:

1. What regions must it cover?

LTE-M and NB-IoT band support is regional and operator-specific. A “global” module on paper does not mean it works on every operator. Build a regional matrix: each target country × each target operator × bands required. Most projects need fewer regions than the brief suggests; cut the matrix to what is actually shipping in year one.

2. LTE-M, NB-IoT, or both?

• LTE-M for products that move (asset tracking, vehicles) or need lower latency. Better signal in dense urban areas.
• NB-IoT for static, low-data devices (utility meters, sensors). Better deep indoor penetration; cheaper data plans in many regions.
• Both if your single SKU has to cover varied deployment profiles. Costs more in module BOM and certification, but reduces SKU sprawl.

3. SIM architecture

The choice matters more in 2026 than it did in 2022:

• Physical SIM — cheapest and simplest, but locks you into one operator per device.
• eSIM (eUICC) — over-the-air SIM provisioning, multi-operator profiles. Now standard in serious IoT deployments.
• iSIM — SIM functionality integrated into the modem itself. Saves space and BOM, but availability is module-specific. Lock-in to the module vendor’s eSIM platform is the trade-off.

For products shipping to more than two countries, eSIM is the right call. iSIM is a technically clean choice when supported but worth understanding the operator ecosystem before committing.

4. Application architecture

Two common patterns:

• Module + separate MCU: a Quectel or u-blox module plus a discrete STM32 or ESP32 running your application. Clean separation, easy debugging. Higher BOM, more board area.
• Module-as-MCU: Nordic nRF9151 or Quectel BG770A running your application on the module’s built-in CPU. Smaller BOM, less RAM/flash to play with.

For products with significant application logic (sensor fusion, on-device AI, complex state machines), module + separate MCU is usually the right call. For simple telemetry products, module-as-MCU saves real money.

The hidden cost: certification

Carrier and regulatory certification is what makes cellular IoT expensive. A pre-certified module on a pre-certified board with no antenna changes is a 4–6 week PTCRB process. A custom RF design with a custom antenna is 12+ weeks and tens of thousands of dollars across PTCRB, GCF, and operator-specific approvals.

Two practical paths:

• Pre-certified plus reference antenna — fastest to market, slightly higher BOM, simpler certification.
• Discrete module + custom RF — lower BOM at volume, but only worth it past 100k units annually.

For most projects we ship the first path on v1, then evaluate the second on v2 once volume justifies the engineering.

What we typically default to

For a cellular IoT product in 2026:

• Module: u-blox SARA-R510 or Quectel BG770A, depending on cost target
• SIM: eSIM with two-operator profiles, sourced through Soracom, 1NCE, or Onomondo
• Application: separate STM32 or Nordic SoC if logic is non-trivial; on-module if it is simple telemetry
• Antenna: pre-certified embedded antenna with reference layout — a separate antenna design only after volume justifies it

If you are mid-decision on a cellular module — and the module choice locks in your operator strategy for the life of the product — we are happy to look at the brief.