Abstract: : The last decade has witnessed a profound transformation in Earth Observation (EO) through the deep learning revolution, with Artificial Intelligence (AI) being increasingly employed to tackle various EO challenges. However, the unique characteristics of EO data, which are in fine geospatial measurements from diverse sensors, and the intricate nature of underlying physical processes (originating in the Earth system, weather and climate, or human activities), pose significant obstacles. As a result, progress in adopting machine learning techniques varies across applications, and the full potential of these methods remains untapped. This presentation will provide an overview of the past decade’s advancements, current status, and challenges in using machine learning for EO, followed by insights into the next stage of the ongoing machine learning revolution, including foundational models and their implications. It will also delve into the future of EO, exploring the emergence of quantum machine learning and quantum computing for EO, as well as the potential of EO processing in modular high-performance computing (HPC) environments.