Issue link: https://resources.mouser.com/i/1481902
STMicroelectronics 2022 3 W e live in a pivotal moment of human history, defined by the emergence of AI across all industries. Although the growth of AI has been astounding, its adoption is just getting started. AI-driven technology will eventually be as helpful as electricity, leading to safer transportation, unlimited virtual assistance, greater security, fewer human errors, and personalized automation for almost every aspect of life. In fact, AI fuels human innovation itself—which accelerates the discovery and development of solutions that enhance global productivity, environmental sustainability, and quality of life. Centralized "Cloud-AI" is what people encounter with services like GPS mapping and personalized web ads. Such applications are compute-intensive and require high- performance GPU cores and AI accelerators for complex data processing. The corresponding performance metrics are FLOPS (Floating-Point Operations Per Second) and TOPS (Tera Operations Per Second) per watt. Despite its massive global impact, the reality of Cloud-AI is that it's power-hungry, bandwidth- hungry, data-hungry, privacy-challenged, and dominated by giant data companies. For these reasons, many other companies have struggled to create customer value with successful production deployments. But this situation is rapidly changing as AI migrates from cloud servers to billions of "smart devices" at the network edge. This is called distributed "Edge-AI", the next AI frontier, where AI algorithms are directly executed on embedded microcontrollers next to users and data sources for snap decision making and failure anticipation. This means that machine learning models can now be deployed on miniature Internet of Things (IoT) and 5G devices, autonomous vehicles, and home appliances without cloud dependency. STMicroelectronics is a leading supplier of STM32 microcontrollers and IoT technologies, and we are committed to enabling Edge-AI democratization and differentiation. Our goal is to accelerate mainstream Edge-AI adoption with machine learning methods, tools and IPs that simplify the data science for engineers. NanoEdge™ AI Studio, for example, allows manufacturers to deploy self-learning electric motors that can predict problems and compensate for aging magnets. This eBook discusses the emergence and importance of embedded machine learning, its potential for customer value creation, and what's coming next. ST also invites you to learn about its Edge AI focus with intelligent sensors, STM32 micro technologies, and the proven STM32Cube.AI and NanoEdge™ AI Studio platforms. ■ Local Machine Learning for responsive AI at the Edge Marc Dupaquier, STMicroelectronics Figure 7 1518 4375 2026 2025 2024 Edge Al Market Forecast 2023 2022 Edge Al Hardware Market (Million USD) CAGR = 18% 2021 0 1000 2000 3000 4000 5000 6000 7000 1834 3708 3145 1257 1040 2666 2259 861 713 1915 Edge Al Software Market (Million USD) CAGR = 20.8%