SemiconductorsSTMicroelectronics unifies STM32 microcontroller platform for cloud and physical AI
STMicroelectronics has introduced a unified STM32 microcontroller platform, new automotive and power chips, and updated development tools for cloud and physical AI applications.
I
IT Magazine Staff
SemiconductorsIn brief: STMicroelectronics unveiled a unified STM32 microcontroller platform and new chip lines at its STM32 Summit 2026 keynote on 3 November 2026, followed by an announcement on 4 November. Built on Arm Cortex-M cores used in over 100 billion devices, the expanded portfolio targets industrial, automotive, cloud AI, and physical AI systems with shared software and development tools.
STMicroelectronics introduced a unified STM32 microcontroller (MCU) platform alongside new automotive, mixed-signal, and neural-accelerated chips at its STM32 Summit 2026 keynote on 3 November 2026, followed by an announcement on 4 November 2026. The semiconductor maker is combining architectures, intellectual property, and software tools across industrial, automotive, and robotics applications to serve growing cloud AI and physical AI markets.
Unified platform and Arm collaboration
Building on more than 20 years of STM32 development, the unified platform enables engineering teams to reuse software, hardware architectures, and manufacturing investments across product generations rather than starting again for each product. According to Maria Heriz, executive vice president of STMicroelectronics' embedded processing subgroup, nearly 2 million unique developers use the STM32 ecosystem, with active users rising by around 20% over the past three years. Remi El-Ouazzane, president of the company's Microcontrollers, Digital ICs and RF Products Group, said the industry needs less fragmentation and a stronger focus on converging MCU technologies around a common foundation to maximise software reuse.
The platform relies on Arm Cortex-M cores, which are used in more than 100 billion devices, alongside the STM32Cube software ecosystem. As a lead partner with early access to Arm cores, STMicroelectronics is helping define the future Cortex-M processor roadmap and works with over 100 companies, including Cadence, in Arm's Total Design programme for physical AI, announced in September. Future STM32 devices will be manufactured using STMicroelectronics' 40-nm flash and 18-nm fully depleted silicon-on-insulator (FD-SOI) technology with embedded phase-change memory, supplemented by embedded non-volatile memory (eNVM) technologies from TSMC.
New automotive and power microcontrollers
Alongside its existing Stellar automotive MCU line, STMicroelectronics unveiled the STM32A family for high-safety systems assessed under ISO 26262 Automotive Safety Integrity Levels (ASILs). The entry-level STM32A4 offers up to 0.5 MB of memory with QM and ASIL-B support for local control nodes and safety systems. The mid-range STM32A5 provides up to 4 MB of memory and ASIL-B support for vehicle body and comfort functions as well as industrial automation, robotics, and energy systems. The STM32A9 supports up to 12 MB of memory, Ethernet connectivity, and a split-lock architecture that can prioritise ASIL-D safety or provide twice the ASIL-B performance for multi-domain real-time control.
For digital power and motor control, the STM32G5 mixed-signal MCU series is in production now starting at $1.034 for 10,000-unit orders. Using an Arm Cortex-M33 core running at up to 280 MHz with 64 Kbytes to 512 Kbytes of flash memory, it delivers twice the compute performance of the STM32G4 series. It introduces a 12-channel high-resolution timer to manage silicon carbide and gallium nitride power supplies as AI data centre rack power approaches 1 MW and peak facility demand could exceed 2 GW by 2030. The series also features a programmable logic array called PLAy, Arm TrustZone, and post-quantum cryptography support, targeting Platform Security Architecture Level 3 certification. Taiwan-based Delta Electronics and Chinese robotics company Ruiyan Intelligent have adopted the chip.
Edge AI chips and development tools
STMicroelectronics also introduced the STM32H7P, combining a Cortex-M7 core with its proprietary Neural-ART neural processing unit delivering 85 giga operations per second (GOPS). In early evaluations, a solar inverter customer used a single MCU to manage arc-fault circuit breakers across 20 panels instead of four. On the software side, the company began rolling out redesigned edge AI tools at the end of October, including Model Zoo, Model Factory, and Dataset Factory, having supported over 260,000 active AI projects and more than 400 pre-validated models over the past 12 months. Additionally, STM32Cube for agentic AI uses a model context protocol (MCP) server to give AI coding agents access to documentation, reference code, and review tools alongside the STM32 Sidekick support tool.
What happens next
The first three STM32A microcontrollers are currently sampling to customers, with broader availability scheduled to start with the STM32A9 in the first half of 2027. Mass production for the STM32H7P is planned for early 2027, and dual-core products will join the STM32G5 series in 2027. Before then, following a proof of concept with NVIDIA on the STM32H7, STMicroelectronics will release Holoscan Sensor Bridge software as a native STM32Cube component in the coming months to bring Ethernet-based sensor and actuator connectivity across its portfolio.
Sources: EE Times, STMicroelectronics, The Elec, Unite.AI
Fact-checked against the linked sources before publication. Spotted an error? See our Corrections Policy.
Spotted an error? Email corrections@itmagazineme.com with the article link.
