nRF54LM20B Ultra-Low-Power Multiprotocol Wireless SoC with Integrated NPU

The Nordic Semiconductor nRF54LM20B, a member of the nRF54L series, is a high-memory, ultra-low-power wireless system on chip (SoC) intended for advanced embedded and connected devices that require deterministic low-power operation and on-device intelligence. It integrates an application MCU, a RISC-V coprocessor, and the Nordic Axon neural processing unit (NPU) on a single device, reducing external processing and power overhead.

This device provides 2 MB of non-volatile memory and 512 KB of RAM, the largest memory configuration in the nRF54L family, enabling complex protocol stacks such as Matter, Bluetooth® Mesh, and Thread alongside application code and edge AI workloads. The integrated 2.4 GHz radio supports Bluetooth LE, IEEE 802.15.4-based protocols, and proprietary modes, with data rates up to 4 Mbps for low-latency links.

The on-chip Axon NPU accelerates inference tasks such as classification and pattern detection, delivering significantly higher energy efficiency than CPU-only execution. Combined with deep sleep modes down to sub-microamp current levels and hardware-based security isolation, the nRF54LM20B is designed for long-life, battery-powered systems that require secure connectivity and local data processing.

Features

• Integrated Axon NPU for efficient on-device AI inference and signal classification
• 2 MB NVM and 512 KB RAM, supporting large firmware images and multiprotocol stacks
• Multiprotocol 2.4 GHz radio supporting Bluetooth LE, Thread, Matter, Zigbee®, and proprietary modes
• Ultra-low-power operation with sleep currents starting below 1 µA and optimized RX/TX current consumption
• Hardware security, including TrustZone® isolation, tamper detection, and cryptographic acceleration with side channel protection

Applications

• Advanced AI designs
• Battery-powered smart home and building automation devices using Matter, Thread, or Bluetooth Mesh
• Industrial wireless sensors requiring local anomaly detection or classification at the edge
• Asset tracking and condition monitoring with low-latency wireless links and long operational life
• Wearable and portable devices requiring secure connectivity and low-power signal processing
• Embedded systems combining local AI inference and multiprotocol wireless connectivity without external processors