Arduino UNO Q: Two Processors, One Board, €39
Qualcomm bought Arduino and put one of its Dragonwing SoCs into the new UNO Q alongside a dedicated STM32 microcontroller. The result is a board that runs full Linux and real-time hardware control simultaneously – at a €39 price point. Here is what it means for IoT.
What is the Arduino UNO Q?
The Arduino UNO Q is the first board in the UNO family to include a Linux-capable application processor alongside a traditional microcontroller. The Q stands for Qualcomm, which acquired Arduino in 2024 and has put its Dragonwing QRB2210 SoC at the heart of the new board. It sits in the standard UNO form factor – same footprint, same shield pinout – but it is a fundamentally different type of device from any UNO that came before it.
There are two processors running simultaneously. The Qualcomm Dragonwing handles the heavyweight work: running a full Debian Linux distribution, AI model inference, image processing, web connectivity, and Python scripts. The STM32U585 microcontroller handles exactly what Arduino has always handled: real-time digital and analog I/O, sensor reading, actuator control, and anything that needs deterministic timing. The two talk to each other over an internal bridge, so a Python script on the Linux side can trigger a motor controller on the STM32 side without any external wiring.
The short version: plug a keyboard, mouse and monitor into the USB-C port and the UNO Q works as a small Linux desktop. Disconnect them and it falls back to its microcontroller role, reading sensors and driving hardware exactly as any Arduino would. It is genuinely both things at once.
Key specifications
One notable change from previous UNO boards is the removal of the DC barrel jack. Power comes via USB-C at 5V up to 3A, or via the VIN pin for 7-24V supplies. The USB-C port also doubles as a display output – connect a USB-C dock and you get HDMI, USB peripherals and power from a single cable.
How the dual-processor architecture works
The architecture is more nuanced than simply bolting a Raspberry Pi to an Arduino. The Qualcomm side runs Debian Linux with full package management, so you can install Python, Node.js, Docker, or any Linux application. The STM32 side runs Zephyr OS and executes Arduino sketches in the normal way – loop(), setup(), digitalRead(), analogWrite(), all of it works exactly as expected.
Arduino App Lab is the new development environment that ties both sides together. It lets engineers write and deploy applications that mix microcontroller sketches, Python scripts, and AI models in a single workflow, with each component running on the appropriate processor.
Qualcomm Dragonwing QRB2210
Running: Debian Linux, Python, Docker, AI models
Connected to: USB-C dock, Wi-Fi, Bluetooth, internet
↔ internal bridge ↔
STM32U585 microcontroller
Running: Arduino sketches via Zephyr OS
Connected to: GPIO pins, sensors, actuators, shields
The practical implication is that you can do things that previously required two separate boards and wiring between them. A Python script can read sensor data from the STM32, process it with a TensorFlow Lite model on the Qualcomm side, and push results to an MQTT broker over Wi-Fi – all within a single €39 board.
UNO Q versus UNO R4 versus Raspberry Pi 5
| Feature | Arduino UNO R4 WiFi | Arduino UNO Q | Raspberry Pi 5 |
|---|---|---|---|
| Main processor | Renesas RA4M1 (32-bit, 48 MHz) | Qualcomm QRB2210 (quad-core, 2.0 GHz) | Broadcom BCM2712 (quad Cortex-A76, 2.4 GHz) |
| Dedicated MCU | Same chip handles everything | STM32U585 (dedicated real-time) | None – GPIO via Linux only |
| Operating system | None (bare metal) | Debian Linux + Zephyr on MCU | Raspberry Pi OS (Linux) |
| RAM | 256KB SRAM | 2GB or 4GB LPDDR4 | 4GB or 8GB LPDDR4X |
| Arduino shield compatibility | Full | Full (same pinout) | None |
| Real-time I/O | Excellent | Excellent (dedicated STM32) | Limited (kernel scheduling) |
| AI inference | Very limited | Capable (Qualcomm NPU) | Capable (external accelerator needed) |
| 5V GPIO compatibility | Yes | 3.3V (Qualcomm platform) | 3.3V |
| Price | ~€25 | €39 | ~€55-70 |
| Best for | Simple sensor/actuator projects | Edge AI + hardware control | General compute, camera, display |
Shield compatibility note: The UNO Q GPIO runs at 3.3V, not 5V. Existing 5V Arduino shields may not work correctly. If your project depends on 5V shield compatibility, the UNO R4 Minima or R4 WiFi remains the better choice.
Where the UNO Q fits in IoT deployments
For most industrial IoT deployments the UNO Q is not a direct replacement for a cellular router or an industrial gateway. It lacks the certified RF hardware, ruggedised enclosure, wide operating temperature range, and DIN rail form factor that field deployments typically need. But there are several scenarios where it slots in usefully.
Edge AI prototyping
Running TensorFlow Lite or ONNX models on data from local sensors without a cloud round-trip. The Qualcomm NPU handles inference while the STM32 manages the sensor interface.
Rapid gateway prototyping
Building a prototype MQTT gateway or protocol bridge before committing to industrial hardware. Run your Modbus or serial polling on the STM32, push results to a broker via Linux networking.
Lab and test benches
Hardware-in-the-loop testing where you need a Linux environment to generate test data and a microcontroller to simulate real-time responses. One board replaces two.
Educational IoT platforms
Teaching cloud connectivity, AI, and hardware control on the same device. The UNO form factor means existing teaching materials mostly still apply.
Low-volume embedded products
Custom controllers for small production runs where the economics of a bespoke PCB do not stack up. The UNO Q gives you a capable compute platform at a reasonable BOM cost.
Smart home and building automation
Local processing with direct hardware control – running Home Assistant or a custom automation engine on Linux while the STM32 handles the relay and sensor I/O.
The Qualcomm acquisition context
Qualcomm completed its acquisition of Arduino in 2024. The UNO Q is the first hardware product to come from that relationship. From Qualcomm’s perspective the move makes strategic sense: Arduino gives them a route into the maker and education market with a well-understood platform, and embedding Dragonwing silicon into a €39 board builds an ecosystem around hardware that would otherwise only appear in industrial modules and reference designs.
For Arduino users the picture is more nuanced. Qualcomm brings serious compute capability and a strong NPU story, but the shift to a 3.3V Qualcomm platform and the move away from simple bare-metal operation changes what the UNO is. The R4 family still exists and is the right choice for anyone who wants a classic Arduino experience with modern 32-bit performance. The UNO Q is a different product category sitting between a microcontroller and a single-board computer.
What about the Raspberry Pi 6?
The UNO Q arrives at an interesting moment for the SBC market. Raspberry Pi 6 has been pushed to early 2028 at the earliest, largely driven by DRAM cost inflation. That leaves a gap in the market for capable, affordable edge compute boards. The UNO Q does not fill that gap in the same way – it is not a general-purpose Linux computer in the way the Pi is – but it does offer something the Pi cannot: genuine real-time hardware control alongside application-layer compute on a single board, at a price point that makes sense for experimentation and low-volume production.
For industrial IoT deployments that need certified cellular hardware, DIN rail mounting and a wide operating temperature range, the choice still comes down to programmable industrial routers and purpose-built edge gateways rather than development boards. The UNO Q is a prototyping and development platform, not a field-deployable industrial device in the same category as a Teltonika RUTC or a Robustel EG5200.
Summary
The Arduino UNO Q is a genuinely interesting board. The combination of a real-time microcontroller and a Linux-capable application processor on a single €39 PCB is something the maker and embedded development communities have wanted for years. It makes the most sense for AI-at-the-edge prototyping, gateway development, and educational platforms where you need Linux convenience and real-time hardware control in the same device.
For production industrial IoT deployments the limitations are real – no industrial certifications, 3.3V only GPIO, consumer-grade operating temperature range, no cellular modem. But as a development platform and a glimpse of where low-cost edge compute is heading, it is worth paying attention to.