Google DeepMind has launched AlphaEarth Foundations, an advanced AI model described as a 'virtual satellite' that integrates petabytes of diverse Earth observation data into high-resolution, 10-meter global maps. This model compresses vast satellite and other geospatial data into compact, analysis-ready summaries, enabling consistent and rapid monitoring of terrestrial land and coastal waters. AlphaEarth Foundations supports tracking environmental changes such as deforestation, coastal shifts, climate variations, urban development, and ecosystem dynamics with unprecedented precision. The AI system generates precomputed global datasets dating back to 2017 without requiring atmospheric correction or cloud masking, facilitating applications in climate forecasting, flood and wildfire prediction, and urban planning. Alongside AlphaEarth, Google has expanded its mapping efforts by adding millions of previously unmapped buildings across Africa to Google Maps, addressing digital infrastructure gaps on the continent. Separately, NASA introduced Galileo, an open-source multimodal AI model that combines satellite, radar, climate, and map data to enhance Earth observation tasks such as crop mapping and flood detection. Additionally, Google released MLE-STAR, an AI agent aimed at automating complex machine learning research tasks typically requiring expert human intervention.