ADVERTISEMENT
Advertisement
AI

STMicroelectronics Launches STM32G5 MCUs for AI Datacenter Power and Robotics

Listen to this story

ⓘ AI NARRATED
0:00 / 0:00

STMicroelectronics has introduced the STM32G5 series of mixed-signal microcontrollers for digital power and motor control in AI datacenters, robotics, and other applications. Built around an Arm Cortex-M33 core running at 280MHz, the MCUs deliver twice the compute performance of the STM32G4 series.

The company cited growing energy needs in AI datacenters, where increasing GPU-cluster performance is driving rack power toward 1 MW and the largest datacenters could demand over 2 GW of peak power by 2030. It also described humanoid robotics as requiring simultaneous multi-axis motor control, high current delivery and fast response, with mechanisms such as a robotic hand potentially using as many as 20 motors.

ADVERTISEMENT
Advertisement

Your details go straight to the advertiser. EE Herald does not keep a copy.

Hi-res PR image

The STM32G5 adds new and upgraded peripherals for power supplies and electric motors. Its enhanced high-resolution timer can support algorithms that dynamically reconfigure power-supply units as server workloads change, and the company says its timing capabilities can handle higher switching frequencies in power supplies using silicon carbide (SiC) and gallium nitride (GaN) semiconductors. The series also includes a position encoder and a programmable logic array (PLAy) for implementing custom functions in hardware.

Key features include a Cortex-M33 core rated at 420 DMIPS/1136 CoreMark @ 280MHz; I/D cache and routine booster hardware accelerators; motor-control timers; a 12-bit analog-to-digital converter (ADC), operational amplifier, 12-bit digital-to-analog converter (DAC) and ultra-fast comparators; and an enhanced 12-channel high-resolution timer, which ST describes as the first of its kind in an MCU.

ADVERTISEMENT
Advertisement

Get a quote

Pricing and lead times, direct from suppliers.

Your details go straight to the advertiser. EE Herald does not keep a copy.

Security features combine the Cortex-M33's Arm TrustZone IP with STM32Trust capabilities, including an immutable Root of Trust (STiRoT) for secure boot and secure firmware update. The MCUs also include a cryptographic engine supporting current algorithms and ST's Post-Quantum Cryptography (PQC) libraries, along with software protection, secure storage, tamper detection, side-channel attack resistance and life cycle security management.

Patrick Aidoune, general manager of the General Purpose and Automotive Microcontrollers Division at STMicroelectronics, said the series addresses the energy demands of AI datacenters, more sophisticated humanoid robots and increasingly complex motor-control systems. He said the compute performance is designed to support physical AI with improvements in sustainability, safety and security.

The STM32G5 series is supported by STM32CubeMX2 and the STM32Cube hardware abstraction layer (HAL2), as well as digital power and motor-control development resources. The MCUs are in production, with on-chip Flash from 64Kbytes to 512Kbytes and package options from QFN32 to LQFP128. Prices start at $1.0340 for orders of 10,000 pieces. Product information is available at www.st.com/stm32g5.

E

EEHerald News Desk


More from AI →