#Norway #Edge AI #Nuremberg #Nordic Semiconductor #nRF54LM20B

Nordic Semiconductor Strengthens Its Edge AI Offerings

Nordic Semiconductor has made significant strides in promoting energy-efficient intelligence at the edge with the recent launch of the nRF54LM20B System-on-Chip (SoC). This innovative hardware introduces a Neural Processing Unit (NPU) designed to facilitate real-time artificial intelligence tasks while maintaining low power consumption. This move marks a pivotal advancement for developers and manufacturers focusing on battery-powered devices.

Innovations in Performance

The nRF54LM20B comes equipped with a powerful integrated NPU that boosts the speed at which TensorFlow Lite-class models operate, achieving performance up to 15 times faster compared to traditional execution on an Arm Cortex CPU. Not only does this enhanced performance enable complex sensor data processing, but it also achieves an impressive eight-fold increase in energy efficiency in comparison to competing edge AI solutions.

Øyvind Strøm, EVP Short-Range at Nordic Semiconductor, highlighted that this generation of edge AI capabilities radically alters how battery-operated devices perceive and interact with the world. By embedding advanced intelligence directly onto these devices, Nordic eliminates the need for cloud processing, optimizing both power efficiency and system complexity.

Practical Applications

The NPU-enabled nRF54LM20B caters to a wide array of battery-operated devices such as wearables, smart home gadgets, medical sensors, and trackers. These devices can now efficiently perform tasks like activity recognition, sound and presence detection, and anomaly identification in real-time without draining energy reserves.

Streamlined Edge AI Development

Recognizing the common hurdles developers face when incorporating AI features into devices, Nordic Semiconductor has refined its edge AI framework. This includes:

• - Streamlined Model Creation: Developers can now effortlessly build and customize edge AI models for on-device execution. By uploading specific datasets and selecting the model architecture, crafting tailored AI models has never been easier.
• - Wake-Word Generation: The new SoC allows developers to produce a wake-word model from just a single phrase. This advancement eliminates the extensive audio dataset requirements previously needed for voice-activated functionalities, simplifying the development of smart speakers and other voice-controlled devices.
• - Ecosystem Support: Nordic's comprehensive online model generation tool provides developers with flexibility to create models optimized for either NPU acceleration or efficient CPU execution. This alleviates integration issues among different models and tools, while still supporting various applications like audio classification and environmental sensing.

Availability and Next Steps

The new nRF54LM20B development kit is available through Nordic's distribution partners. Sample requests for the SoC itself are welcome as volume production is set to commence in Q2 2026.

Additionally, Nordic Semiconductor will showcase its latest advancements at the upcoming Embedded World 2026 event in Nuremberg, further solidifying its commitment to pioneering low-power edge AI solutions. This demonstration is positioned to exhibit the profound transformations the nRF54LM20B can facilitate across various industries.

Conclusion

As the landscape of IoT continues to evolve, Nordic Semiconductor's introduction of the nRF54LM20B SoC not only cements its position as a leader in low-power wireless technology but also democratizes access to edge AI capabilities for developers and manufacturers alike. With these tools, businesses can create smarter, more responsive products that balance performance with energy conservation. It's a monumental step forward in paving the way for a new generation of intelligent devices.