Wi-Fi SoCs pair low power with edge AI
Nordic reports that its nRF71 series of Wi-Fi SoCs delivers up to 20 times lower standby power consumption than the nRF70 series of Wi-Fi companion ICs. Both devices draw an average connected current of 14 µA at 2.4 GHz and 8 µA at 5 GHz when operating at DTIM 10. The nRF7120 offers dual-band Wi-Fi 6, while the nRF7120E adds tri-band Wi-Fi 6E and an Axon NPU for hardware-accelerated edge AI. Each device supports Bluetooth LE, Thread, and Matter and features 4 MB of on-chip non-volatile memory and 1 MB of user-accessible RAM.
The nRF71 series’ ultra-low power consumption is expected to enable new battery-powered Wi-Fi designs for smart home, wearable, and industrial IoT applications. Its integrated NPU allows AI inference to run directly on the chip, eliminating the need for an external AI processor while reducing latency and power consumption. Nordic cites keyword spotting, anomaly detection, and image classification as practical edge AI applications for battery-powered devices based on the nRF71 series.
With the Axon NPU, machine-learning models run 10× to 17× faster than on the 256-MHz Cortex-M33 alone. The nRF71 series can also execute ultra-small AI models trained on developers’ own data using Nordic’s Neuton software, enabling edge AI inference directly on the processor without NPU acceleration.
The nRF71 series is now sampling with selected customers.
The post Wi-Fi SoCs pair low power with edge AI appeared first on EDN.
