Nordic Semiconductor, founded in 1983 in Trondheim, Norway, develops ultra-low-power wireless connectivity solutions. Originally Nordic VLSI, the company launched 433 MHz products in 1998 and rebranded in 2003, focusing on 2.4 GHz wireless devices. Nordic manufactures the nRF54- and nRF5340-series SoCs featuring dual Arm Cortex-M33 processors and multiprotocol transceivers. Recently acquiring Neuton.AI and Atlazo, Nordic integrates TinyML capabilities into its SoCs, enabling sub-5kB AI models for edge applications in IoT, wearables, and industrial devices.
(Source: Nordic)
Nordic Semiconductor was founded in 1983 in Trondheim, Norway. The company was established by four postgraduates from the Norwegian University of Science and Technology, who initially focused on designing mixed-signal ASICs and was initially called Nordic VLSI (NVLSI).
The company develops wireless connectivity solutions for edge computing applications and manufactures the nRF54-series SoCs that operate with minimal power consumption requirements.
NVLSI launched its first wireless standard products in the 433 MHz ISM band in 1998. The company rebranded to Nordic Semiconductor in 2003, adopting the Nordic Semi moniker to emphasize its focus on ultra-low-power wireless devices. Nordic Semiconductor released its first wireless devices operating at 2.4 GHz in the same year. Since 2003, Nordic Semiconductor has concentrated exclusively on wireless products for the 2.4 GHz band, achieving widespread adoption across numerous branded consumer electronic products throughout various market segments.
In November 2019, Nordic Semiconductor introduced a flagship SoC containing dual-core Arm Cortex-M33 processors and a multi-protocol radio stack, and in August 2023, it acquired the IP portfolio of San Diego-headquartered artificial intelligence and machine learning company Atlazo.
Figure 1. Nordic Semiconductor’s SoC. (Source: Nordic)
The nRF5340 ASIC is a wireless, ultra-low-power SoC, integrating two fully programmable Arm Cortex-M33 processors, security features, and a multi-protocol 2.4 GHz transceiver. The transceiver supports Bluetooth Low Energy, ANT, and IEEE 802.15.4 for Thread and ZigBee protocols. It also allows the implementation of proprietary 2.4 GHz protocols.
The two Arm Cortex-M33 processors share the power, clock, and peripheral architecture with Nordic Semiconductor nRF51, nRF52, and nRF91 series of SoCs, ensuring minimal porting efforts. The application core is a full-featured Arm Cortex-M33 processor including DSP instructions and FPU, running at up to 128 MHz with 1MB of flash and 512 kB of RAM. The option to run the application processor at 64 MHz allows the CPU to increase energy efficiency. The network core is an Arm Cortex-M33 processor with a reduced feature set, designed for ultra-low-power operation.
While it does not produce dedicated AI processors in the traditional sense, Nordic has been actively integrating AI and ML capabilities into its SoCs and expanding its offerings through acquisitions and partnerships.
The company has focused on enabling AI at the edge by combining its ultra-low-power wireless SoCs (such as the nRF54 and nRF53 series) with TinyML frameworks. This allows for scalable, high-performance AI on resource-constrained devices, such as wearables and IoT sensors.
Nordic Semiconductor is not producing stand-alone AI processors but is integrating AI and ML capabilities into its existing ultra-low-power wireless SoCs, making them suitable for edge AI applications in IoT and wearable devices.
Recently, the company acquired the IP and core technology assets of Neuton.AI, a pioneer in automated TinyML solutions, to enhance its AI capabilities for edge devices. Additionally, it acquired Atlazo, a US-based company specializing in always-on AI processors and energy management for tiny edge devices, further strengthening its position in low-power AI/ML.
Neuton.AI develops automated machine learning platforms that generate neural network models for resource-constrained embedded systems. The acquisition integrates these technologies to enable on-device AI processing capabilities.
Neutron.AI’s platform generates machine learning models that consume less than 5kBof memory storage. The automated system creates models without requiring manual parameter tuning or specialized data science expertise. The platform supports deployment across 8-bit, 16-bit, and 32-bit microcontroller architectures. The neural network framework constructs models without predefined architectural templates, enabling deployment on devices with limited computational resources.
The technology addresses power consumption constraints in edge computing applications. Traditional machine learning inference requires data transmission to remote processing centers, consuming additional power for wireless communication. Local processing reduces communication overhead and extends battery operation duration. The approach targets consumer electronics, healthcare monitoring devices, and industrial sensing applications.
Vegard Wollan, CEO and president at Nordic Semiconductor, states that the acquisition enables developers to construct AI-powered devices with reduced power consumption and form factor requirements. The integration targets applications requiring continuous operation with battery power constraints.
The acquisition addresses market demand for edge intelligence capabilities. TinyML chipset shipments project growth to $5.9 billion by 2030, according to market analysis. Applications include predictive maintenance systems, health monitoring devices, process automation, gesture recognition interfaces, wearable consumer electronics, and Internet of Things sensors.
Neuton.AI’s automated platform addresses traditional barriers to embedded machine learning deployment. Manual model optimization requires specialized knowledge and iterative development processes. The automated approach reduces development time and enables broader adoption across engineering teams without dedicated machine learning expertise.
The integration targets specific application domains where local processing provides operational advantages. Healthcare monitoring benefits from reduced latency and improved privacy protection through local data processing. Industrial applications gain from reduced communication dependencies and improved reliability. Consumer devices benefit from extended battery life and responsive user interactions.
The acquisition positions Nordic to address growing demand for intelligent edge devices across multiple market segments. The integration provides developers with tools to implement machine learning capabilities without specialized AI expertise, while maintaining the power efficiency requirements of battery-operated devices.
The AI processor market is a fast-growing, evolving one, and Jon Peddie Research covers it in the JPR AI Processor Market Development Report, a supply-side report series that establishes the AI Processing market size by shipments, value, segment type, installed base, and investment from the current quarter back to Q1’14 for a historical perspective.
WHAT DO YOU THINK? GOOD ENOUGH TO TELL YOUR FRIENDS ABOUT IT? YOU HAVE OUR PERMISSION TO SHARE IT.