Imagine trying to fit a giant encyclopedia into a tiny pocket notebook without losing the important information. This project creates a set of tools that shrinks complex AI brains so they can fit onto small chips inside gadgets. It lets these devices think for themselves instantly without needing to send data to a far-away cloud server.
Companies currently rely on non-European AI tools that are often too large or power-hungry for small devices. This creates a dependency on foreign tech and leads to high latency and security risks when sending data to the cloud.
The AIDGE toolset, which allows users to shrink, optimize, and deploy AI models directly onto embedded hardware chips.
If you are a component manufacturer dealing with high latency in braking systems — this project developed the AIDGE toolset that allows AI to run locally on the chip. This ensures real-time responses and better safety for the driver.
If you are a startup dealing with patient data privacy and battery drain — this project developed optimization methods that reduce model size. This keeps sensitive health data on the device and extends battery life.
If you are a producer dealing with unreliable internet in factories — this project developed a way to deploy AI on custom hardware. This allows machines to detect faults instantly without relying on a cloud connection.
Based on available project data, the platform is described as open-source, implying it is designed for broad accessibility rather than a proprietary paid license.
Yes, the project involves 16 industry partners, including large companies like Thales and STMicroelectronics, to ensure the tools meet industrial requirements for reliability and safety.
The project aims to provide a sovereign, open-source platform to end dependence on American and Chinese tools, though specific license types are not detailed in the text.
It provides tools to convert and deploy AI applications across a large number of industrial or custom hardware architectures, including components from NanoXplore and STMicroelectronics.
The project runs from June 2023 to November 2026, with the first stable version of the AIDGE tool released in April 2024.
The consortium is heavily weighted toward commercial application, with a 57% industry ratio consisting of 16 companies. This includes a strong mix of 9 SMEs and 4 large enterprises (Thales, IFAG, TTTech Auto AG, STMicroelectronics), ensuring that the resulting software is grounded in real-world hardware constraints and market needs across 5 European nations.
Contact the Commissariat à l'énergie atomique et aux énergies alternatives (CEA) in France.
Contact us to identify compatible European hardware partners for your AI deployment.
