Nordic Semiconductor nRF71 Series brings edge AI to battery Wi-Fi SoCs
Nordic Semiconductor has announced the nRF71 Series, a family of ultra-low-power Systems-on-Chip (SoCs) designed to bring Wi-Fi, Bluetooth Low Energy, Thread and on-device AI inference together on a single die. The Oslo-based chipmaker says the first two parts, the nRF7120 and nRF7120E, consume up to twenty times less standby power than the existing nRF70 Series, a figure it positions as the key enabler for battery-powered Wi-Fi devices in smart home, wearables and industrial IoT markets. The nRF71 Series is currently sampling with selected customers.
The standby power claim is stated by Nordic and has not been independently benchmarked. Both devices include 4 MB of on-chip non-volatile memory and 1 MB of user-accessible RAM. The nRF7120 supports dual-band Wi-Fi 6, while the nRF7120E adds tri-band Wi-Fi 6E alongside Nordic's Axon Neural Processing Unit (NPU) for hardware-accelerated inference. Nordic says the Axon NPU runs machine learning models between ten and seventeen times faster than the chip's 256 MHz Arm Cortex-M33 processor when used alone, enabling practical use cases such as keyword spotting, anomaly detection and image classification without an external AI co-processor.
Technical design
Two independent radios, each with separate RF and processing blocks, allow the nRF71 Series to maintain a Wi-Fi cloud connection and simultaneously monitor Bluetooth LE or Thread traffic. That simultaneous receive capability matters for Matter-ecosystem devices, such as smart locks, security panels and Thread border routers, which often need to bridge multiple wireless protocols without handoff latency. Nordic's Neuton software lets developers train and deploy ultra-small AI models from their own data, and over-the-air model updates are supported via the company's nRF Cloud service, addressing the lifecycle management challenge that has complicated edge AI deployments.
Kjetil Holstad, executive vice president of Wi-Fi and power management at Nordic Semiconductor, said: "With the nRF71 Series, Nordic removes that barrier. We are bringing Wi-Fi and edge AI to the same ultra-low-power platform and tools developers already trust for Bluetooth LE, so the next generation of intelligent products can connect to any network, make decisions on the device, and still run for years on a battery."
Market context
The edge AI silicon market has become intensely competitive, with established players including NXP, Silicon Labs, Espressif and Infineon all offering varying combinations of wireless connectivity and ML acceleration for IoT endpoints. Nordic has historically built its franchise on Bluetooth LE, where it commands a strong design-win position among consumer and medical device OEMs. Extending that franchise to Wi-Fi and on-device inference is a strategic push into a higher-value tier of the IoT chip market.
The Matter smart-home standard, backed by Apple, Google, Amazon and the Connectivity Standards Alliance, has accelerated demand for chips that handle multiple wireless protocols simultaneously. SoCs that can serve as Thread border routers while running Wi-Fi cloud connectivity are increasingly a baseline requirement for Matter certification, and Nordic's dual-radio architecture directly addresses that requirement.
Regulatory tailwinds are also relevant. The EU's Cyber Resilience Act, which will apply to connected products from late 2027, requires manufacturers to ensure devices receive security and software updates throughout their supported lifetime. Over-the-air model and firmware update capability, such as the nRF Cloud pipeline Nordic describes, is likely to become a procurement checkbox rather than a differentiator as that regulation bites.
Nordic will demonstrate the nRF71 Series at Electronica in Munich, in Hall C6 at booth 159. No pricing or volume availability dates have been disclosed beyond the sampling stage.