Abstract: The semantic EO data cube combines an EO data cube with an AI-based inference engine by integrating a computer-vision approach to infer new information from EO data in an explainable way. Our approach uses semantic enrichment of optical EO images to obtain data-derived information, i.e., an initial interpretation, and makes them directly available and accessible for further analysis within the cube. The architecture is based on an expert system, in which domain knowledge can be encoded in semantic models (knowledgebase) and applied to the image data as well as data-derived information (factbase) using a graphical inference engine. This approach offers new and innovative functions for semantic, content-based image search and image analysis in user-defined regions and time intervals using a newly developed semantic querying language, which does not require programming skills. The world's first prototype of this approach is the Sen2Cube.at data and information cube covering all of Austria, including all available Sentinel-2 images since Sentinel-2A's launch in 2015. Thus, the prototype was successfully scaled to a national data cube for Austria. This presentation will focus on the architecture of this approach and demonstrate the transferability to other regions (Syria) and sensors (AVHRR).