14.30 – 14.55
As edge AI workloads continue to increase, system designers face growing challenges related to power consumption, thermal limits, and system complexity when using traditional GPU-based approaches. This session explores the transition toward energy-efficient NPUs that are purpose-built for on-device AI inference.
You will learn how modern edge AI architectures can deliver high performance per watt while maintaining accurate, high‑precision inference, making them well suited for deployment in constrained environments. The session highlights how DEEPX technology enables efficient local AI processing without relying on power-hungry accelerators.
Practical examples will be discussed across a range of applications, including smart cameras, robotics, industrial automation, and IoT systems, showing how AI workloads can be accelerated while staying within strict power and thermal budgets.
Speaker: Michaël Uyttersprot – manager AI/ML & Embedded Vision for Avnet Silica