/dq/media/media_files/2025/01/22/Ghd10JxUSeJtzy4m0Vaf.jpg)
STMicroelectronics is making edge AI a reality with high-performance MCUs. Sridhar Ethiraj, Group Manager, Technical Marketing India, General Purpose MCU Division, STM, said that the explosion of AI-enabled devices is accelerating the inference shift from cloud to the tiny edge. Tiny-edge centric is enabled by a different class of hardware and software.
We need to look at the benefits of tiny edge. These are: ultra-low latency for real-time apps, reduced data transmission to generate meaningful information, enhanced privacy and security in cloud, sustainable on energy with low data and no power, and lower cost of inference to enable new class of operations.
We are opening a new range of embedded apps. We cam upgrade existing devices with AI-based services. These include arc fault detection, battery management, predictive maintenance, etc. We can also reduce BoM by shifting MPUs to MCUs for apps such as people detection, speech recognition, etc. We offer advanced services without the cloud.
We are increasing the adoption of edge AI solutions. More inference tasks will be handled locally, which will help reduce the strain on global power grid. We have significant growth in developer engagement. We also have a roadmap for challenging roadmaps. Neural=ART accelerator is a dedicated ST proprietary neural processing unit. We can accelerate edge AI adoption, and enable energy efficiency. We have introduced the Neural-ART 1 MCU.
STM has introduced the STM32N6 MCU with AI acceleration. We are pushing this into two categories -- general purpose and AI. We are enabling high-performance edge AI on MCUs. This is elevating both graphics and multimedia experiences. STM32N6 is seamlessly integrated into the STM32 ecosystem. It is supported by ST Edge AI Suite tools, and is compatible with TouchGFX packages for graphics. It provides 600x ML performance uplift. STM32H7 is an existing MCU. This new MCU exceeds performance by 600x. You can benefit from MPU-like experience in industrial and consumer apps.
STM32N6 features include a dedicated embedded neural processing unit (NPU), with 600 GOPS NPU and three TOPS/w power consumption. Arm Cortex-M55 core, with 1280 DMIPS / 3360 CoreMark. 4.2Mb of embedded RAM for real-time data processing and multitasking. Enhanced computer vision capabilities, with enhanced parallel and MIPI CSI-2 with camera module IP, and dedicated image processor (ISP). Extended multimedia capabilities, with 2.5D graphics accelerator, H.264 encoder, and JPEG encoder / decoder. Finally, advanced security features, with Arm Trustzone for Cortex-M55 core and NPU. Targeted certifications are from SESIP3 and PSA L3.
STM has development ecosystem for lower barrier to entry and faster project successes. We also have the STM32 Cube.AI to integrate into apps. We are also giving the STEdge AI developer cloud to test and download models.
We are innovating across product segments. Examples are, drones, smart industry, smart homes, smart farming, robots, automotives after market, personal electronics, personal healthcare, smart buildings, etc.
Meta-Bounds is enhancing wearable technology with STM32N6 MCU. This is the most powerful MCU by ST. It is supported by wireless connectivity solutions. We are boosting edge AI adoption.