Infineon's PSoC Edge MCUs Enhanced with Nvidia AI Toolkit
Munich, Thursday, 20 March 2025.
Infineon integrates Nvidia’s TAO toolkit into its PSoC Edge MCUs, enabling streamlined development and deployment of AI models for smart applications like home automation and wearables.
Revolutionary Edge AI Integration
At Embedded World 2025, Infineon Technologies unveiled a significant advancement in edge computing by announcing the integration of Nvidia’s TAO (Train, Adapt, and Optimize) toolkit with their PSoC Edge microcontroller family [1]. This integration addresses a crucial challenge in modern electronics: implementing sophisticated AI development workflows at the edge while maintaining efficiency and performance [2].
Technical Specifications and Capabilities
The PSoC Edge microcontroller family showcases impressive technical specifications, built on a 22 nm ultra-low-power embedded RRAM process. The platform features hardware-assisted machine learning acceleration and incorporates a secured enclave with a trusted execution environment [1]. The system architecture includes Arm Cortex-M55 processors and Ethos-U55 microNPU, specifically designed to enhance energy-efficient machine learning capabilities at the edge [2].
AI Development Streamlined
The TAO toolkit empowers developers with comprehensive AI capabilities, supporting various functions including image classification, object detection, segmentation, optical character recognition, body pose estimation, action recognition, and sensor fusion [1]. According to Steve Tateosian, Senior Vice President IoT Compute & Wireless at Infineon, this integration ‘significantly speeds up time-to-market for ML enabled applications in industrial automation, medical, automotive, and smart IoT solutions’ [2].
Market Impact and Future Prospects
The collaboration between Infineon and Nvidia represents a strategic move in the edge computing landscape. As confirmed by Deepu Talla, Vice President of Robotics and Edge Computing at Nvidia, ‘NVIDIA TAO brings the latest advances in computer vision models and fine-tuning workflows to the far edge’ [2]. The integration promises to enhance system determinism, privacy, and security while reducing latency, system power, and costs for AI-enabled solutions [2].