Nordic Semiconductor strengthens nRF54L Series leadership in ultra-low-power edge AI

Nordic Semiconductor, a global leader in low‑power wireless connectivity solutions, today announces a major step forward in its edge AI strategy, expanding its next‑generation, ultra-low-power portfolio with new cutting-edge capabilities and the broad availability of its first Neural Processing Unit (NPU) enabled nRF54LM20B System-on-Chip (SoC).

High-performance acceleration for real-time intelligence

The large-memory nRF54LM20B’s integrated NPU accelerates TensorFlow Lite–class models up to 15 times faster and with significantly lower energy consumption than execution on the Arm Cortex CPU, and delivers up to 7 times higher performance and 8 times better energy efficiency than the closest competing edge AI solution - making high‑rate sensor data, audio, and event‑driven edge AI workloads practical on a tiny battery.

“This new generation of edge AI-enabled capabilities fundamentally transforms what small, battery-powered devices can perceive and interpret in the real world,” says Øyvind Strøm, EVP Short-Range at Nordic Semiconductor. “By bringing powerful intelligence directly onto the device – without compromising power efficiency, latency, or system complexity – we are enabling an entirely new class of products that are significantly smarter, more responsive, and dramatically more energy-efficient than anything previously possible.”

By simplifying how edge AI models are created, optimized, and deployed, Nordic makes it easier than ever for developers to add meaningful intelligence to their products without increasing power budgets or development overhead.

Enabling real-time intelligence in everyday devices

Nordic is strengthening its edge AI offering by combining advanced hardware acceleration, flexible model support, and intuitive developer workflows to make on-device intelligence practical for mainstream products. The new NPU-enabled nRF54LM20B provides the performance needed for real-time intelligence in battery-powered wearables, smart home and audio devices, industrial and medical sensors, and trackers. This enables accurate activity detection, natural sound and presence response, instant anomaly identification, and movement interpretation without cloud processing.

This new generation of edge AI-enabled capabilities fundamentally transforms what small, battery-powered devices can perceive and interpret in the real world

Øyvind Strøm, EVP Short-Range at Nordic Semiconductor

Simplifying edge AI development

In addition to the new hardware acceleration, Nordic has expanded its ultra-low-power edge AI solution with features that remove common barriers to adding intelligence to a wide range of battery-powered, connected IoT products:

Streamlined model creation and optimization:
Developers can now build edge AI models for on‑device execution by uploading custom datasets and specifying the model architecture to generate classification or regression models. This simplifies the design of edge AI architectures from scratch, making them ideal for wearables, industrial sensing, and smart home devices.
Wake‑word creation from a single text input:
This allows developers to generate a custom wake‑word model from a single written phrase (e.g., “Hello Nordic”), without collecting or labeling thousands of audio samples. This solves the high cost and difficulty of assembling voice datasets, enabling voice‑activated controls, smart speakers, wearables, remote controls, and battery‑powered audio accessories.
Broader ecosystem support:
Models from Nordic’s new online model generation tool – Nordic Edge AI Lab, can be created for accelerated inference on the NPU or efficient execution on the CPU - giving developers flexibility across performance and power requirements. This solves integration friction between models, tools, and embedded firmware while enabling audio classification, anomaly detection, advanced IMU processing, predictive maintenance, environmental sensing, and signal processing.

Together, these updates reduce the effort typically required to bring AI features to embedded devices and dramatically shorten the time from concept to working prototype.

Availability

The nRF54LM20 DK is now available through Nordic’s distribution partners. For nRF54LM20B SoC samples, please contact Nordic Sales. Volume production is expected to begin in Q2 2026.