High-Performance Acceleration for Real-Time Intelligence
The nRF54LM20B’s integrated NPU, coupled with large memory capacity, processes TensorFlow Lite class AI models up to 15 times faster than execution on an Arm® Cortex® CPU, while also substantially reducing energy consumption. This new SoC achieves up to 7 times the performance and 8 times the energy efficiency compared to the nearest competing edge AI solutions. This capability makes it practical to implement edge AI workloads such as high-frequency sensor data processing, audio analysis, and event-driven AI processing in small, battery-powered devices.
Øyvind Strøm, EVP of Short Range Devices at Nordic Semiconductor, stated that this new generation of edge AI capabilities fundamentally transforms how small, battery-powered devices can perceive and understand the real world. By implementing powerful intelligence directly on devices without compromising power efficiency, latency, or system complexity, the nRF54LM20B enables a new category of smarter, more responsive, and energy-efficient products.
Real-Time AI for Everyday Devices
Nordic Semiconductor is enhancing its Edge AI solutions by combining advanced hardware acceleration, flexible model compatibility, and intuitive development workflows. This approach aims to make on-device AI practical for mainstream products. The NPU-equipped nRF54LM20B is designed to enable real-time AI in battery-powered devices such as wearables, smart home equipment, audio devices, industrial sensors, medical sensors, and tracking devices.
Simplified Edge AI Development
Beyond hardware acceleration, Nordic Semiconductor has expanded its ultra-low power edge AI solutions with features designed to overcome common barriers to integrating intelligence into battery-powered connected IoT products. These features include:
-
Simplified Model Creation and Optimization: Developers can now build edge AI models by uploading their own datasets and specifying model architectures. This simplifies the creation of classification and regression models, significantly streamlining the process of designing edge AI architectures from scratch. This is ideal for applications in wearables, industrial sensing, and smart home devices.
-
Wake Word Creation from Single Text Input: Developers can generate custom wake word models by simply entering a single text phrase (e.g., “Hello Nordic”), eliminating the need to collect and label thousands of audio samples. This significantly reduces the cost and effort associated with building audio datasets, making it easier to develop voice-controlled devices. Applications include voice-operated devices, smart speakers, wearables, remote controls, and battery-powered audio accessories.
-
Expanded Ecosystem Support: Models created with Nordic’s new online model generation tool, Nordic Edge AI Lab, can be configured for either NPU-accelerated inference or efficient execution on the CPU. This offers flexibility in meeting both performance and power consumption requirements. It also addresses challenges associated with integrating AI models, development tools, and embedded firmware, enabling applications such as audio classification, anomaly detection, advanced IMU processing, predictive maintenance, environmental sensing, and signal processing.
Availability
The nRF54LM20 DK is currently available through Nordic Semiconductor’s distributors. Samples of the nRF54LM20B SoC can be requested by contacting Nordic Semiconductor’s domestic sales representatives. Mass production is planned for Q2 2026.
About Nordic Semiconductor
Nordic Semiconductor is a global leader in low-power wireless connectivity solutions, offering a comprehensive platform that spans hardware, software, development tools, and cloud-based lifecycle services. This platform simplifies and accelerates the development of connected products while ensuring reliable performance throughout their lifecycle. Founded in 1983 and headquartered in Norway, Nordic Semiconductor employs approximately 1,450 people worldwide. The company has pioneered Bluetooth® Low Energy (Bluetooth LE) and expanded its wireless solutions to include Cellular IoT, Wi-Fi, Matter, Thread, Zigbee, DECT NR+, and NTN/satellite communication. In 2025, Nordic enhanced its chip-to-cloud value proposition with the acquisition of Memfault, adding advanced device monitoring capabilities and cloud-based observability to its comprehensive wireless product portfolio. Nordic’s technologies and solutions enable secure, scalable, and energy-efficient connected products in consumer, healthcare, and industrial sectors, supporting the development of a smarter, more connected world. For more information, please visit https://www.nordicsemi.com.
