IBM Releases TerraStackAI to Bridge Geospatial AI Tooling Gap

IBM has released TerraStackAI, a new open-source technology stack designed to make it easier to work with Earth and space observation AI models. The system addresses a critical gap: while foundation models for satellite data have advanced rapidly, the tooling to use them effectively has remained fragmented and inaccessible.

"Foundation models like IBM-NASA's Prithvi-EO and IBM-ESA's TerraMind enable unprecedented multimodal representations of the Earth system," IBM noted. "However, a critical gap remains: the tooling required to effectively use these models is fragmented, complex, and inaccessible to many who need to use them."

TerraStackAI consists of four integrated components:

TerraKit: Handles data preparation, connecting to satellite sources including Copernicus Sentinel-1 (radar), Sentinel-2 (optical), and NASA's Harmonized Landsat Sentinel-2 (HLS). It abstracts away format conversions, projections, and preprocessing complexity.

TerraTorch: A modular framework for fine-tuning and evaluating foundation models, built on PyTorch Lightning and TorchGeo.

Iterate: Automates hyperparameter search using Bayesian optimization, integrated with MLFlow and Ray to parallelize experiments.

Geospatial Studio: A production platform with both no-code interfaces and programmatic APIs for deploying AI services at scale.

The system was designed for use cases including extreme weather risk monitoring, disaster response, precision agriculture, and solar activity forecasting—domains that rely on turning petabytes of satellite data into actionable insights.

TerraStackAI is now available via GitHub, with the Geospatial Studio integration announced alongside Red Hat.