On November 23, Arm announced the launch of the Arm® Cortex®-M52 processor designed specifically for artificial intelligence Internet of Things (AIoT) applications, enhancing digital signal processing (DSP) and machine learning (ML) performance while effectively avoiding the cost overhead associated with dedicated DSP and ML accelerators. The Cortex-M52 will fully unleash the potential of ML in embedded computing solutions with advantages over current market price points.

Currently, generative artificial intelligence and large language models (LLM) are receiving significant attention in the industry. However, most people are still unaware that artificial intelligence has already been widely deployed in many embedded devices and is influencing various applications in homes, cities, and industries. This is referred to as the artificial intelligence Internet of Things, and the Arm architecture is the cornerstone of it all. Artificial intelligence is crucial for understanding data and achieving seamless interaction between the physical and digital worlds.

Paul Williamson, Senior Vice President and General Manager of Arm’s IoT Division, stated: “With the evolution of AIoT technology, intelligence at the device end is being deployed at the lowest cost into smaller, more cost-sensitive devices that are often battery-powered, and due to the reduced dependency on the cloud, they possess higher privacy and reliability. As the shipment of AI-driven IoT devices continues to grow, our partners’ demand for ML capabilities and simplified development processes is also increasing, providing them with the flexibility to innovate and scale rapidly.”

Bringing AI capabilities to a wider range of IoT devices

The Cortex-M52 incorporates Arm Helium™ technology, which provides significant performance improvements for DSP and ML applications in small, low-power embedded devices, enabling the deployment of more compute-intensive ML inference algorithms at the endpoint without the need for a dedicated NPU. Helium technology has been successfully deployed in edge products at remote network locations, and the launch of the Cortex-M52 will help Arm partners apply Helium technology to devices with lower costs and more constrained power.

By extending Helium technology to the new Cortex-M processor, Arm is significantly changing the matrix of microcontrollers and DSP computing in small embedded devices. The Cortex-M52 offers a simplified migration path from Cortex-M33 and Cortex-M4, catering to a wide range of AIoT application needs, enabling richer user interfaces (UI), voice, and visual experiences, such as automotive and industrial control, predictive maintenance, and wearable sensor fusion. The Cortex-M52 provides scalable flexibility across a range of performance points and configurations while offering DSP capabilities without the need for a separate processing unit, thereby saving chip area and cost.

Delivering ideal performance and excellent cost choices while ensuring robust security and reliability

The Cortex-M52 elevates the efficiency of the Cortex-M series based on the Armv8.1-M architecture (including Cortex-M55 and Cortex-M85) to a new height and marks an important milestone in introducing ML capabilities to microcontrollers. Among all Cortex-M processors supporting Helium technology, it achieves the lowest area and power consumption, providing chip partners seeking to balance performance and cost with new options. Developers will also benefit from the enhanced ML and DSP performance, with the ML performance of the Cortex-M52 being up to 5.6 times better and DSP performance up to 2.7 times better compared to previous generations of Cortex-M.

Security is a key concern for various devices, especially during mass shipments. The Cortex-M52 implements the latest security extensions of the Armv8.1-M architecture, including pointer authentication and branch target identification (PACBTI) and Arm TrustZone® technology, to better defend against software threats. Additionally, the Cortex-M52 aids in accelerating the launch of PSA Certified Level 2 certified chips, thereby facilitating the next generation of PSA Certified devices. The latest Armv8.1-M cores (including Cortex-M55 and Cortex-M85) provide enhanced functional safety features, which are crucial for many automotive and industrial control applications. Compared to the previous generation Cortex-M deployed in these applications, the extended security packages and features offered by the Cortex-M52 help partners achieve FuSa certification more quickly and easily.

Simplifying AI development for ultra-small endpoint devices

Traditional embedded developers need to master the mathematical, data analysis, toolchain expertise, and programming skills required for artificial intelligence. Developer engagement is crucial to increase the shipment of AIoT devices. Through the Cortex-M52, Arm provides the key features and functionalities required for modern development processes.

In the past, achieving the ML and DSP performance provided by the Cortex-M52 often required a combination of CPU, DSP, and NPU. However, this meant that once the hardware was ready, developers had to use three different toolchains, compilers, and debuggers to write, debug, and optimize the code for the chip. Now, Arm offers an industry-standard and user-friendly Cortex-M product portfolio that allows artificial intelligence to be implemented on a single toolchain and a single validated architecture, ensuring that traditional DSP and ML workloads can achieve a consistent development process, while specific integrations and optimizations for leading ML frameworks will ensure developers have a seamless development experience that fully leverages the excellent performance of Cortex-M.

The Cortex-M52 is fully compatible with the Cortex-M55 and Cortex-M85 software, allowing developers to benefit from the continuously evolving software and tool ecosystem around Helium, and make the most of the free software libraries and rich knowledge bases from the Arm partner ecosystem. To help simplify and accelerate the IoT and embedded development process, the Cortex-M52 can also be accessed in the form of Arm virtual hardware, a cloud-based solution that allows software development to start before the chip is ready.

Arm is the cornerstone for achieving AI across all AIoT scenarios

The Cortex-M52 was previously known as “Star-MC2” in the Chinese market and is a collaborative product developed by Arm and the engineering team at Anmou Technology.

Arm is the cornerstone of AIoT. Arm and its partners have recognized the market demand—to provide the necessary DSP and ML computing performance for low-power embedded applications at a lower cost and with better accessibility. The Cortex-M52 CPU was born to bring a higher level of artificial intelligence reasoning performance to ultra-small devices, helping the industry further expand the deployment scale of IoT devices.