ST Unveils Unified STM32 MCU Platform

STMicroelectronics has announced a unified STM32 microcontroller (MCU) platform at its 2026 STM32 Summit. Spanning across automotive, industrial, and robotics applications, the unified platform enables customers to reuse architectures, intellectual property (IP), software, and tools. ST also expanded its STM32 family with new and scalable mixed-signal and high-performance MCUs for software-defined systems and physical AI, along with new and enhanced edge AI tools and agentic coding.

Described as the industry's first unified MCU platform across industrial, automotive, and robotics applications, it delivers long-term scalability, faster time to market, lower investment risk, and lower total cost of ownership. The platform is designed around three pillars: portfolio consistency, a common shared ecosystem, and shared process technology.

The portfolio consistency centers around the Arm Cortex-M cores, which makes it easier to implement software and IP into new designs. ST also announced that it will be a lead partner with Arm on future Cortex-M generations. The company also recently joined Arms' Total Design program for physical AI.

The common MCU software ecosystem includes the STM32Cube and ST's Edge AI Suite, which is expected to support two million developers by the end of 2026 with active users growing about 25% a year since 2023. ST has more than 400 optimized AI models and the STM32 Sidekick AI assistant. It is adding agentic AI features to the STM32Cube.

The shared process technology consists of ST's 40-nm Flash and 18-nm FD-SOI technology with embedded phase-change memory, along with TSMC's embedded non-volatile memory technologies.

In a product briefing, Remi El-Ouazzane, president, STMicroelectronics' Microcontrollers, Digital ICs and RF Products Group, explained there is a technology shift happening in the industry that is driving the adoption of shared technology foundations. The technology convergence whereby automotive, industrial, and robotic systems face similar requirements in AI, security, safety, connectivity, sensing, and control, he said.

One of the most important outcomes [of this technology convergence] is the emergence of next-generation systems that can perceive, decide, and act autonomously in the real world, and "physical AI is the embodiment of that convergence mixing within the same framework a combination of different classes of MCUs and MCU subsystem," he said.

Sharing the same connectivity, real-time control, safety, security, and software lifecycle management constraints is the technology foundation that is increasingly becoming shared across automotive, industrial, and robotics domains, El-Ouazzane said. "As this foundation becomes more common they become more reusable, and it's becoming easier for developers to scale and deploy across applications."

The challenge for those customers is shifting from selecting a single device to finding a way to reuse software, tools, and architecture, and generating expertise across multiple products and evolving requirements, he added.

El-Ouazzane described it as a value shift from the individual MCU to the broad platform that surrounds it, which is ST's strategy. Instead of looking at automotive, industrial, and robotics as separate domains, ST is building on a shared technology foundation, which is the unified MCU platform approach.

However, he noted that certification and qualification remain market specific, but the underlying technology is becoming more aligned, enabling the embedded community to reuse common foundations and know-how, while adapting them to each domain's requirements faster.

A unified MCU platform approach

The unified STM32 platform brings together common technologies, software development, and ecosystem investments. This brings several advantages including long-term scalability, faster time to market, lower investment risk, and lower total cost of ownership.

"The goal is to help our customers reuse more of what they already know and already have been building for many years in order for them to reuse and scale much farther, whether it is for industrial, automotive, or robotics," said Maria Heriz, STMicroelectronics' executive vice president, Embedded Processing Sub-Group.

Customers will be able to build on common MCU cores and reuse the software development experience and ecosystem investments across different product generations and application domains, she said.

They can also achieve a much faster time to market by reusing existing software assets, tools, and workflows to reduce the development effort, she continued, and lower the investment risk by building on a mature software ecosystem-the STM32Cube, and ST's long track record of quality and reliability.

"It's also going to reduce the total cost of ownership because the engineering teams can reuse tools and expertise instead of starting from scratch on every new product," Heriz said.

The platform is built on three main pillars-portfolio consistency and continuity, a common software ecosystem foundation, and shared process technologies across domains, she explained.

Customers can rely on familiar architectural principles based on Cortex-M cores and shared technology blocks such as security, AI, and safety that they can find across ST products, she said.

The second pillar is about leveraging a common ecosystem foundation, the STM32Cube, including the ST Edge AI Suite, which helps close to two million developers to reuse, tools, software assets, and development know-how, according to Heriz. "It provides more than 400 STM32 AI models, the largest MCU optimized library of its kind."

Users also can leverage recent AI-powered advances. In addition to the recently introduced STM32Sidekick tool that adds 24/7 digital assistant functionality, ST announced its STM32Cube for Agentic AI framework to support workflows incorporating AI agents in STM32 MCU-based projects. It provides STM32-specific information directly from ST to help AI agents deliver coding accuracy and minimize token costs. It is in beta now and will be released soon.

STM32 platform expansion

Building on more than 20 years of development, the STM32 platform consists of over 4,000 part numbers for automotive, industrial, and wireless applications, including high performance, mainstream, mixed-signal, ultra-low-power, and wireless MCUs across this common software ecosystem, Heriz said.

New with the unified STM32 platform is the launch of the STM32A MCUs, extending the STM32 platform into high-safety and automotive applications. ST introduced the first three products.

"The STM32A will bring higher-safety requirements and automotive-grade capabilities into a platform approach that can increasingly serve on industrial and robotics use cases," Heriz said. "It's all about scalability on one side and continuity on the other."

The automotive-grade family consists of the STM32A4 for entry-level automotive real-time nodes and safety systems with QM/ASIL-B/crypto support, and the STM32A5 for mid-range safety-critical applications for automotive, industrial automation, and robotics with ASIL-B safety and enhanced security (post-quantum cryptography hardware security module [PQC HSM]).

At the high end is the STM32A9 safety-critical multi-domain controller, where real-time control, low latency, Ethernet connectivity, and deterministic communication becomes essential, Heriz said.

All three devices are sampling. Additional STM32A products are planned for 2027.

Alongside the STM32A, ST also introduced the STM32G5 mixed-signal and STM32 H7P MCUs.

The STM32G5 targets real-time power control for AI data centers and better motor control for robotics. It offers double the performance of the previous generation STM32G4, supports PQC, and is supported by the new STM32CubeMX2 ecosystem, said Patrick Aidoune, STMicroelectronics' group vice-president and general manager, general purpose and automotive microcontrollers.

The STM32CubeMX2 targets newer STM32 MCUs, starting with the STM32C5 family.

The STM32G5 MCUs, built with the Arm Cortex-M33 core running at 280 MHz, offers new and upgraded peripherals for higher speed and precision for controlling equipment such as power supplies and electric motors. The STM32G5 is ST's first MCU with an enhanced 12-channel high-resolution timer and it introduces the new PLAy programmable logic array, which allows developers to implement special custom functions in custom hardware reducing the need for external glue logic.

For security, ST combines the Cortex-M33's Arm TrustZone IP with proprietary STM32Trust features that include an immutable Root of Trust for secure boot and secure firmware update, an advanced cryptographic engine that supports the latest algorithms, including ST's PQC libraries, software protection, secure storage for sensitive data such as cryptographic keys, tamper detection, side-channel attack resistance, and support for life cycle security management.

The STM32H7P MCUs for edge AI add hardware acceleration for digital signal processing and edge AI workloads with a 2773-CoreMark Arm Cortex-M7 to handle digital power, motor control, condition monitoring, and other mission-critical industrial applications. The architecture includes ST's hardware signal processor (HSP) and Neural-ART neural processing unit (NPU).

The HSP is a cost-optimized and energy-efficient DSP accelerator for matrix, vector, and digital signal processing operations and helps free up Cortex-M7 bandwidth and can reduce the need for a dedicated external DSP, according to ST.

The Neural-ART NPU frees edge-AI workloads from software constraints, delivering up to 85 GOPS. Demos show a 40-times faster inference than software running ST's AI Arc Fault Circuit Interrupter (AFCI) model for solar-inverter safety. One solar customer can now monitor 20 panels for arc faults with a single MCU, Aidoune said.

Other demos show the anomaly detection model runs 20 times faster, and the AI battery-management system (BMS) model runs 15 times faster.

Software support includes the Edge AI Suite, X-CUBE-MCSDK-2 motor-control software tools, and the upgraded STM32Cube ecosystem. The NUCLEO-H7P5II evaluation board and STM32H7P5IDK Discovery kit will be available in January 2027.

New and redesigned edge AI tools
In addition to the new STM32Cube for Agentic AI, ST has redesigned the workflow in the ST Edge AI Suite with four new tools-Model Explorer, Model Factory, Dataset Factory, and Model Bench-to streamline the development of embedded machine-learning models on the STM32 MCUs.

Model Explorer helps developers find the right pre-trained AI models for the STM32 MCUs, Model Factory helps customize and optimize these models, Dataset Factory organizes and cleans the data, and Model Bench allows customers to benchmark the different models on the ST board, Aidoune explained. The new tools will be available in late November.

ST has supported over 260,000 projects through its edge AI tools over the past 12 months, according to Aidoune.

The ST Edge AI Suite connects users with the broader STM32 ecosystem with over 4,000 commercial part numbers. The STM32Cube ecosystem provides tools to select and configure MCUs, provides software libraries and code examples, and lets users work within their preferred integrated development environment. It also connects with evaluation hardware, from low-cost STM32 Nucleo boards to full-featured Discovery kits, and expansion cards such as wireless connectivity, vision, time-of-flight, IoT sensing, and motor drives.

The post ST Unveils Unified STM32 MCU Platform appeared first on Embedded.