Google DeepMind AlphaEarth
AlphaEarth Foundations is a Google DeepMind model that distils petabytes of multi-sensor Earth observation into annual 10m, 64-dimension embeddings. The resulting Satellite Embedding dataset launched in July 2025 and is available in Google Earth Engine.
Google DeepMind AlphaEarthInside Google DeepMind AlphaEarth
AlphaEarth Foundations assimilates optical, thermal, radar, elevation, climate and even text inputs into a single 64-dimension embedding per 10m pixel, acting like a virtual satellite.
The derived Satellite Embedding dataset is annual, globally complete over land and shallow water, and reported at over 1.4 trillion embedding footprints per year.
Distributed through the Earth Engine Data Catalog and Google Cloud Storage, the embeddings let teams classify land cover and change from sparse labels, which is valuable in low-label tropical regions; the dataset is open even though the model itself is not.
A ready embedding layer that turns a handful of field labels into project-scale classification, well suited to label-scarce tropical NbS sites.
Recent developments
- Jul 2025Google DeepMind launched AlphaEarth Foundations and the annual Satellite Embedding dataset in Earth Engine.
Using Google DeepMind AlphaEarth in a project
The provider’s own site holds its current capabilities, access terms and documentation. Read them before selecting the tool or planning an integration.