Nagold - 8 June 2026 - Edge AI systems bring intelligence directly to the point of action: data remains at the machine, camera or sensor, decisions are made in milliseconds - without detours via the cloud. This ensures greater speed, higher reliability and lower operating costs. As an AI system integrator, endrich has expanded its AI portfolio: the DX-M1M edge AI accelerator from DeepX brings up to 25 TOPS of AI computing power with just 3 watts of power consumption directly into the device, where it handles inference for image, video and sensor data in real time. In the compact M.2-2242 format (M+B), it relieves the CPU and GPU, reduces data traffic and makes edge AI applications economical even in compact embedded systems.
"The DX-M1M offers exactly the capabilities required for industrial edge AI processing," says Antonio Rodas, Chief Technology Officer at endrich. The Edge AI accelerator has a power-saving 2 GB LPDDR4x RAM. Data is exchanged with the host via a PCIe Gen3 x2 interface. The module utilises two data lines and thus conserves resources in embedded SoCs that only provide a limited number of PCIe lanes. The DX-M1M supports x86 and ARM platforms as well as common frameworks such as ONNX and PyTorch. It fits seamlessly into existing embedded Linux environments.
Thanks to its compact size and low power consumption, the solution can also be used economically in large quantities, for example in robots, smart factories and vision systems. Other areas of application include advanced driver assistance systems (ADAS), AI dashcams and drones as well as edge servers and AIoT devices.
For endrich, DeepX is a strategic technology partner with whom the company is further expanding its portfolio as an AI solution provider. endrich offers more than just hardware: "With our partner network, we bring together AI components, embedded platforms, sensor technology, software and integration expertise to create a complete solution. We support our customers throughout the entire Edge AI application process," explains Antonio Rodas.
Published on 08.06.2026
