STMicroelectronics has officially begun mass production of the STM32MP23x series—comprising the STM32MP235, STM32MP233, and STM32MP231—following the earlier launch of the STM32MP25 line. This new family targets cost-sensitive industrial applications that can benefit from features like an integrated Neural Processing Unit (NPU), heterogeneous multicore architecture (featuring a dual-core Cortex-A35 and a Cortex-M33), real-time operating systems beyond Linux, and robust networking interfaces supporting Time-Sensitive Networking (TSN).

By offering a 16-bit DDR4/LPDDR4/DDR3L memory controller and a dedicated H.264 video decoder—without the overhead of video encoding—the STM32MP23x enables efficient machine learning at the edge. It’s particularly well-suited for applications that require intelligent video analysis but operate within limited budgets.

STM32MP23x: Focused Power Where It Matters

Modern industrial upgrades can be complex, involving tightly coordinated machines and sensors under precise controller management. As machine learning expands into areas such as predictive maintenance, defect detection, and process optimization, there's growing demand for microprocessors that can integrate seamlessly with existing Linux-based infrastructure and engineer familiarity with UNIX-like environments.The STM32MP23x fits that need with dual 1.5GHz Cortex-A35 cores for Linux-based tasks and a Cortex-M33 core for ultra-low-power real-time control, mirroring the flexibility of the more advanced STM32MP25 series—but in a more cost-optimized design. This balance enables developers to avoid delaying upgrades due to budget constraints, thereby enabling faster iteration and smoother transitions in industrial systems.

Unlocking Smarter Edge AI Everywhere

As edge AI continues to gain traction, simplifying development is key. Many businesses recognize the benefits—like lower operational costs and improved quality control—but cost remains a barrier. Take, for example, an automated bottling line: sensors on motors can enable predictive maintenance, reducing downtime, while AI algorithms detect visual defects, increasing customer satisfaction.But the power of AI goes far beyond the NPU alone.

More Than Just a Neural Engine

While the NPU—capable of 0.6 TOPS—is a standout feature, the STM32MP23x also includes a camera interface with Lite-ISP, allowing fast image capture and processing. Its hardware image decoder can handle up to 5MP/s, and it supports rich user interfaces through MIPI DSI and LVDS. Unlike competing solutions that emphasize only NPU performance, STM32MP23x provides a full-stack solution—capturing, processing, and analyzing image data—all with a single MPU, without needing extra companion chips.

This all-in-one integration simplifies deployment in use cases like smart bottling lines, where fast response times and compact hardware are essential.

Tailored for Industrial Challenges

In smart factories, integration is everything. Microprocessors must play multiple roles: powering machine learning, controlling operations, and supporting human-machine interfaces (HMIs). The STM32MP23x is designed with this in mind—offering not just intelligence but also reliability, with features like:

• CAN-FD buses
• SDIO 3.0 support
• USB 2.0 host/device capabilities
• USB Type-C with Power Delivery
• Dual Gigabit Ethernet with TSN

These features, inherited from the STM32MP25 series, provide the determinism and low latency that factory automation demands. TSN allows time-critical data to take priority, supports redundancy for reliability, and helps ensure that industrial networks run with high precision and minimal disruption.