Planetary

Wherobots, the Spatial Intelligence Cloud, enables any data team to analyze data about the physical world faster, at greater scale, and at lower cost compared to traditional solutions. Built by the creators of Apache Sedona, it's a compute lakehouse engine that unifies spatial and non-spatial data, automates data workflows, and runs AI on planetary scale imagery. Spatial data refers to information about places, objects, or activities. Examples include GPS points and tracks, routes, land, road, parcel, crop, and building data, as well as imagery from drones and satellites. This data is fundamental to various industries including aerospace, mobility, ag-tech, insurance, energy, telecommunications, retail, and logistics. In one solution, Wherobots handles these diverse spatial data types and formats, with customers seeing production workloads run up to 20x faster and at lower cost than popular lakehouse engines.

AlphaEarth Foundations is a new AI model from DeepMind that acts like a “virtual satellite” by integrating massive, heterogeneous Earth observation data, optical and radar imagery, 3D laser mapping, climate simulations, and more, into a unified, highly compact embedding for land and coastal areas, enabling consistent, on-demand planetary-scale mapping with far less storage overhead than prior systems. It fuses disparate streams to solve data overload and inconsistency, producing summaries that are 16 times smaller than comparable methods while delivering better accuracy (about 24% lower error in tested tasks) even where labeled data is sparse. Annual embedding collections are published as the satellite embedding dataset in Google Earth Engine and are already being applied by institutions to classify unmapped ecosystems, monitor agricultural and environmental change.