Infineon launches the DEEPCRAFT AI suite, optimised for PSOC Edge microcontrollers, fully unlocking the potential of its edge AI ecosystem

In this interconnected world, Edge AI is redefining how people live, work and interact. To drive this transformation, Infineon Technologies AG (FSE: IFX / OTCQX: IFNNY) has officially launched the DEEPCRAFT AI Suite to expand its Edge AI portfolio. This comprehensive suite of software, tools and solutions, designed to help customers seamlessly integrate AI into their products, is now available. Optimised for Infineon's PSOC Edge microcontrollers, the DEEPCRAFT solution enables developers to fully leverage Infineon's edge AI ecosystem, balancing on-demand performance, energy-efficient design and system-level optimisation. Through this offering, Infineon provides comprehensive support for application designers throughout all stages of edge AI development – from early concept and model development phases to energy-efficient hardware deployment.

Infineon PSOC Edge E84 AI Kit

Steve Tateosian, Senior Vice President and General Manager of the IoT, Consumer and Industrial Microcontrollers Division at Infineon Technologies, stated: "The launch of the DEEPCRAFT AI Kit further expands Infineon's edge AI software ecosystem to unlock the full potential of edge AI. Our edge AI solutions aim to provide a comprehensive suite of software and hardware offerings, such as the DEEPCRAFT AI Kit which incorporates an AI Hub equipped with pre-installed models and audio tuning tools. As the preferred partner in edge AI, we deliver advanced products and solutions that facilitate the seamless integration of AI/machine learning (ML) capabilities into next-generation IoT edge devices. Customers can either develop models from scratch or integrate ready-made models and solutions into their products, thereby accelerating time-to-market."

Infineon's complete edge AI ecosystem

Developers can access all solutions through Infineon's centralised platforms, DEEPCRAFT AI Hub and ModusToolbox, enabling them to contribute their own models or utilise ready-made ones. Thanks to its availability on Infineon hardware, this ecosystem reduces barriers from model development to commercial product integration.

DEEPCRAFT AI Hub: Infineon has launched DEEPCRAFT AI Hub, a one-stop platform for edge AI software. Offering over 50 content resources, including open-source models, Infineon software, tools, solutions, and research cases for industrial, consumer, and automotive applications, it provides inspiration and practical guidance for all stages of AI development.

DEEPCRAFT Studio: Beyond existing support for time-series data such as audio and radar, DEEPCRAFT Studio now extends its capabilities to computer vision. This end-to-end platform facilitates the development of high-performance edge AI and machine learning models. It empowers users to create stable, high-quality models ready for direct deployment in commercial products, with complimentary adaptation for Infineon hardware. This includes training and deploying high-performance computer vision models for object detection using advanced YOLO models – all fully integrated within DEEPCRAFT Studio.

DEEPCRAFT Model Converter: To streamline deployment, developers can utilise the DEEPCRAFT Model Converter to optimise proprietary and open-source models for efficient operation on Infineon hardware. Supporting mainstream AI frameworks such as PyTorch, TFLite, and Keras, this converter enables seamless model execution on PSOC Edge and PSOC 6 microcontrollers.

PSOC Edge Microcontrollers: The PSOC Edge series of Arm® Cortex®-M microcontrollers delivers high-performance, low-power, secure hardware with machine learning acceleration, laying the foundation for next-generation edge applications. Beyond machine learning acceleration, the series incorporates advanced security features including Infineon's Edge Protect Category 4 (EPC4) security framework with PSA Level 2/4 certified iSE, PCI pre-certification, and secure enclaves, enabling end-to-end protection from concept to production. A dedicated 2.5D GPU enables responsive, high-quality graphical interfaces at the edge, delivering photorealistic visuals at significantly lower performance and power costs than traditional 3D processors. PSOC Edge fully supports ModusToolbox, Zephyr®, and the DEEPCRAFT AI suite.

Development of Voice-Controlled Products Based on PSOC Edge

The kit also incorporates DEEPCRAFT voice and audio solutions for developing high-quality voice-controlled products. DEEPCRAFT Audio Enhancement improves speech clarity by eliminating background noise, while DEEPCRAFT Voice Assistant enables natural language interfaces to run locally on edge devices. Both solutions are optimised for the PSOC Edge architecture, utilising the Cortex-M55 dual-core processor paired with Ethos-U55 at up to 400MHz, alongside the Cortex-M33 + NNLite core at up to 200MHz, delivering outstanding performance and energy efficiency for machine learning applications.

On the PSOC Edge, this voice and audio solution enables always-on listening with power consumption below 1mW in an ultra-low latency environment, avoiding repeated wake word prompts and extending battery life. It achieves over 98% detection accuracy in close-range conversation scenarios with minimal false alarms. Compared to traditional approaches, customers can develop AI voice and audio applications nearly twice as fast, significantly reducing development costs.

Beyond low-power always-on listening, PSOC Edge delivers up to 75% faster audio processing than comparable products while consuming approximately half the energy, further enhancing device performance and extending battery life. The combination of AI acceleration and an advanced memory architecture enables the operation of multi-megabyte models such as voiceprint ID—models that exceed the capabilities of modern DSPs. With these capabilities, developers can directly integrate powerful yet energy-efficient voice and audio intelligence systems into their products, accelerating development and bringing smart edge-enabled products to market faster.