Nordic Edge AI

Ultra-low-power edge AI across the industry's broadest range of applications

Your data. Your model. Your edge.

From saving bandwidth and energy to more responsive real-time performance, implementing AI in your embedded applications offers massive benefits beyond the buzzwords. Nordic Semiconductor offers two unique technologies, Neuton models and Axon NPU, exclusively to our customers, to cover the industry's broadest range of devices, applications, and customer needs.

 Neuton models

  

Axon NPU
Custom Neuton models are ultra-tiny edge AI models built from your data using our patented network-growing algorithm, ideal for running edge AI on any Nordic SoC or SiP using its main application core (CPU).    The Axon NPU is our dedicated AI accelerator core, designed to increase the speed and efficiency of TensorFlow Lite models, built into our most capable SoCs. 
 Neuton_Logo_horizontal_Gradient_RGB.png    Axon_Logo_horizontal_Gradient_RGB copy.png
     
     

10x

 10x

< 5 KB 

  

15x 

8x

7x 
Smaller memory footprint than TensorFlow Lite models Faster and more energy efficicent than running TensorFlow Lite models on the CPU  Average memory footprint of custom Neuton models created with our framework    Faster and more energy efficient than running the same TensorFlow Lite model on the CPU  More energy efficient compared to the closest competing product Faster inference compared to the closest competing product
           

 

 

  

About edge AI

Introducing Axon NPU

Introducing custom Neuton models

Why edge AI?

Edge AI is ideal for low-power wireless applications because it enables devices to analyze data locally, reducing the need for constant radio transmission - the biggest energy drain for wireless systems. By processing sensor information directly on the device, you minimize network traffic, cut latency, and maintain functionality even when connectivity is intermittent or unavailable. This local intelligence allows devices to wake less often, transmit only meaningful results, and operate for months or years on small batteries. At the same time, keeping data on-device enhances privacy and security while enabling faster, more reliable decision-making.

What do we mean by "edge AI"?

Edge AI is the concept of running AI algorithms and neural networks on hardware that is not a part of the cloud. This covers a wide range of devices, from edge servers and network infrastructure, through powerful computers, laptops and smartphones, down to the microcontroller inside Nordic's ultra-low-power SoCs and the devices that use them. When we talk about edge AI in Nordic, we're typically only referring to the subset of edge AI that is relevant for Nordic SoCs and SiPs — running AI algorithms and neural networks on low-power wireless devices — bringing edge AI to the very edge. 

Edge AI explanation graphic for webpage

Two unique, complementary technologies

Nordic Semiconductor offers two unique technologies for Edge AI, only available on Nordic hardware, targeting different use-cases and supporting different hardware targets.

Neuton models

Neuton models are ultra-tiny edge AI models, created by our proprietary neural network framework. Their size and efficiency make running edge AI applications on the normal CPU core of any Nordic SoC a viable option for solving a number of AI challenges, based on time-series sensor data like the outputs from accelerometers, IMUs, PPGs, temperature- and electricity-measurement sensors.

Axon NPU

Axon is our integrated AI accelerator, designed to increase the speed and efficiency of TensorFlow Lite models. The Axon NPU will be integrated into select Nordic SoCs and is suited for solving edge AI challenges that rely on higher-rate time-series data, as well as audio and image classification tasks.

Unique self-growing neural networks

Neuton models

Neuton models are ultra-tiny edge AI models, created by our proprietary neural network framework. Their size and efficiency make running edge AI applications on the normal CPU core of any Nordic SoC a viable option for solving a number of AI challenges. Applications based on time-series sensor data like the outputs from accelerometers, IMUs, PPGs, temperature- and electricity-measurement sensors are ideal for Netuon models. Neuton models are grown neuron by neuron to create the most accurate, efficient and smallest network possible. This eliminates the need for manual optimization and user input of network architecture, creating unique neural networks with no unneccessary neurons or connections.  

 

    Neuton_STILL_Network_06.jpg

DevZone blog

Custom Neuton models enable tiny, low-power edge AI on any Nordic SoCs, letting developers build efficient, edge AI models without requiring deep data science knowledge or experience.