How to Pick the Right MCU for Your Next BLE Product

There are roughly 40 BLE-capable microcontrollers on the market right now. You’d think that means 40 real choices. It doesn’t. About 3 chip families account for the vast majority of shipping BLE products, and the gap between picking the right one and the wrong one can mean months of wasted prototyping, a $30K certification surprise, or a supply chain dead end that kills your timeline.

If you’re a founder, PM, or early-stage team making your first embedded hardware decision, this guide is a decision framework, not an encyclopedia. You’ll walk away knowing which BLE chip family fits your product, whether to use a bare SoC (a chip that bundles processor and radio together) or a module (that same chip wrapped with an antenna and pre-certification), and how to avoid the traps that bite people 6 months in.

Start With Your Product, Not the Chip

Before you open a single datasheet, answer 5 questions:

1. What’s your power source? Coin cell, USB-powered, or rechargeable battery?
2. What data are you sending? Periodic sensor readings, audio, or firmware updates?
3. Do you need Wi-Fi or Thread/Zigbee alongside BLE?
4. What’s your target production volume? Hundreds, tens of thousands, or 100K+?
5. How fast do you need a working prototype?

Write your answers down before reading further; they’ll act as your filter for everything below. Coin-cell power immediately favors Nordic, because their sleep currents are measured in single-digit microamps. Need Wi-Fi and BLE on the same chip? That’s Espressif territory. Planning to ship 200 units for an enterprise pilot? A pre-certified module saves you from a $20K RF testing bill.

The Big 3: nRF52 vs ESP32 vs STM32WB for BLE Chip Selection

Three families dominate BLE product design in 2025 and 2026. Here’s what each is actually good at, stripped of marketing language.

Nordic Semiconductor (nRF52 / nRF54 series)

Nordic has long set the standard for BLE silicon. The nRF52840 is probably the single most popular BLE SoC in production wearables, medical devices, and battery-powered sensors. The newer nRF54L15 pushes power efficiency even further.

Strengths: Best-in-class power consumption for BLE, a mature and well-documented SDK (nRF Connect SDK, built on Zephyr RTOS), and a huge ecosystem of pre-certified modules from companies like Raytac, u-blox, and Laird.

Watch out for: No built-in Wi-Fi (there’s a companion chip, the nRF7002, but it’s a separate piece of silicon). The SDK has a steeper initial learning curve than Espressif’s Arduino path.

Espressif (ESP32-C3, ESP32-C6, ESP32-H2)

Espressif built their reputation on cheap Wi-Fi chips, then bolted on solid BLE support. The ESP32-C3 gives you Wi-Fi and BLE 5.0 on a RISC-V core for under $2 in volume. The C6 adds Thread/Zigbee. The H2 drops Wi-Fi but adds 802.15.4 for Matter-ready designs.

Strengths: Lowest barrier to entry by a mile. Arduino and PlatformIO support means your full-stack developer can have something blinking in an afternoon. Aggressive pricing. Enormous community.

Watch out for: Higher power consumption than Nordic when doing BLE-only work, and less proven in regulated or medical contexts. If you don’t need Wi-Fi, that extra radio just costs you power for nothing.

STMicroelectronics (STM32WB55, STM32WBA series)

STMicro’s BLE play targets teams that already live in the STM32 ecosystem. If your product has a beefy STM32 doing motor control or signal processing and you need to tack on BLE, the WB/WBA series lets you stay in STM32CubeIDE with familiar HAL libraries.

Strengths: Natural fit for existing STM32 shops. Strong long-term availability commitments (ST is good about not EOL-ing parts quickly). Solid hardware security features including secure boot and crypto acceleration.

The tradeoff: This family has the smallest BLE-specific community of the three, and fewer off-the-shelf modules. If you’re not already in the STM32 world, there’s little reason to start here just for BLE.

BLE SoC Comparison at a Glance

Why Most Startups Should Start With a Module

A bare SoC is just the chip. You handle the RF layout, antenna matching network, crystal, passive components, and then you pay for FCC/CE/IC testing on your custom RF design. A module packages all of that onto a small PCB with a pre-certified antenna. You solder it down and write your application code.

The cost difference per unit is real. A bare nRF52840 costs roughly $3.50, while a Raytac MDBT50Q module built around it costs $6 to $8. That $3 to $4 delta stings when you’re doing the BOM math.

But here’s what the BOM math misses: custom RF design and certification testing runs $15K to $50K+ depending on complexity and how many regions you’re filing in. At 1,000 units, the module premium costs you $4,000 total. The bare-SoC path saves you that $4,000 on parts but adds roughly $25,000 in engineering and RF testing. Net result: you’re $21,000 worse off than if you’d just used the module.

Rule of thumb: use a module until your volumes justify a custom RF layout. That crossover point is often around 10K+ units, sometimes higher. Plenty of successful products ship on modules permanently. It’s a production-grade decision, not a prototyping crutch.

Real module options worth looking at: Raytac MDBT series (nRF52/54), ESP32-C3-MINI-1 from Espressif directly, u-blox NINA and ANNA series, and Laird Connectivity’s BL series.

If you’re integrating with Hubble Network’s connectivity platform, the device SDK already supports several Nordic and other BLE SoCs, which can simplify your firmware bring-up regardless of whether you’re using a module or bare chip. You can check supported devices to see what’s already been validated.

Certification, Supply Chain, and the Stuff That Bites You Later

Certification is not optional. FCC (US), CE (EU), and IC (Canada) are mandatory for any product with a radio. Pre-certified modules handle the intentional radiator portion, which is the expensive part. You’ll still need to test your complete product for unintentional emissions, but the module dramatically simplifies the process and shrinks the testing bill.

Supply chain should influence your chip decision from day one. The 2021 to 2023 shortage era taught the industry painful lessons. Most BLE SoCs are single-sourced (only one company makes the nRF52840, only Espressif makes the ESP32-C3).

What you can do: design around a module footprint that multiple vendors support. Raytac and u-blox both make nRF52840-based modules with similar pinouts. If one vendor has a 26-week lead time, you might be able to swap to the other with minimal board changes.

Check DigiKey and Mouser stock levels before committing to a platform. Both Nordic and Espressif have strong availability right now, but “right now” changes. A 5-minute stock check can save you from a 6-month production delay.

Picking Your Prototyping Path

Buy a dev board. Seriously, buy two from different families. Spend a weekend with each. The SDK experience will tell you more than any comparison article (including this one).

Nordic: The nRF52840 DK ($40) is the standard. If you want something more beginner-friendly, the Adafruit Feather nRF52840 Express works with Arduino and CircuitPython. For teams going the Zephyr route, there’s a Zephyr reference application that shows a real integration pattern.

Espressif: The ESP32-C3-DevKitM-1 ($8) is absurdly cheap. The Seeed XIAO ESP32-C6 is compact, has USB-C, and costs $5. Hard to beat for a first weekend of prototyping.

STM32: The NUCLEO-WB55RG ($30) integrates with STM32CubeIDE. Best suited for teams who already know the STM32 toolchain.

If you’re a founder or PM who won’t be writing firmware yourself, the dev board exercise still matters. Hand two boards to your engineer or contractor and ask them to evaluate SDK documentation, community support, and build system quality. Their answer will probably make the decision for you.

The Quick Decision Flowchart

Need Wi-Fi + BLE? → ESP32-C3 or C6.

No Wi-Fi needed. Power-critical? (coin cell, tiny battery, multi-year runtime) → Nordic nRF52840 or nRF54L15.

Not power-critical. Already using STM32 in your design? → STM32WB55 or STM32WBA.

None of the above apply? → Default to Nordic nRF52840. It has the strongest overall BLE ecosystem, the most module options, and the deepest professional community.

In every case: start with a module, validate with a dev kit, and let your product requirements drive the decision.

Before You Lock In

This choice isn’t irreversible at the prototype stage. Switching chip families before you’ve committed to a custom PCB costs you a few weeks of firmware rework, not a product redesign. The expensive lock-in happens when you go to production with a custom RF layout around a bare SoC.

Pick a family. Buy the dev board. Build the smallest possible proof of concept that exercises your actual use case: not a blinking LED, but your real sensor data flowing to a phone app. You’ll know pretty quickly whether the platform fits.

Hubble Network enables BLE connectivity from any device directly to satellite—no gateways, no terrestrial infrastructure. See how it works →