Nordic Semiconductor, a specialist in low-power wireless communication solutions, has further enhanced its technological competitiveness by unveiling a new System-on-Chip (SoC) targeting the next-generation edge AI era.

Nordic announced on the 20th that it is expanding its ultra-low power edge artificial intelligence portfolio by unveiling the nRF54LM20B, its first product with an embedded neural network processing unit (NPU).

This new product is characterized by being designed to enable real-time artificial intelligence processing even on battery-based devices.

The NPU integrated into the chip provides up to 15 times faster inference performance compared to existing Arm Cortex CPUs while significantly reducing power consumption.

In particular, compared to competing edge AI solutions, performance is improved by up to 7 times and energy efficiency by up to 8 times, making it suitable for environments requiring high-speed processing such as sensor data, audio analysis, and event detection.

Nordic expects that through these technologies, on-device AI can be easily applied to various general-purpose devices, such as wearable devices, smart home products, industrial and medical sensors, and location tracking devices.

Activity recognition, sound detection, and anomaly analysis without cloud integration It can immediately perform intelligent functions, offering significant advantages in terms of response speed and security.

Development convenience has also been enhanced.

Nordic simplifies the process of creating and optimizing edge AI models, enabling developers to implement classification and regression models without the need for complex architecture designs.

In addition, a wake word model can be generated with just a single sentence input, significantly reducing the time and cost required for voice-based interface development.

Along with this, the newly unveiled 'Nordic Edge AI Lab' supports both NPU-accelerated and CPU-based models, allowing for flexible selection based on performance and power requirements.

This is expected to lead to expansion into various edge AI application fields, such as audio analysis, inertial sensor processing, predictive maintenance, and environmental monitoring.

Nordic Semiconductor's strategy with the nRF54L series is to accelerate the popularization of ultra-low-power edge AI technology and open up a new market where practical AI utilization is possible even in battery-powered devices.