NASA and IBM Introduce Open-Source Lunar AI Foundation Model

NASA and IBM introduced the NASA-IBM Lunar Foundation Model on September 10, 2026, an open-source artificial intelligence system designed to transform how researchers study the Moon. Built using data from several NASA missions—most notably the Lunar Reconnaissance Orbiter (LRO)—the model focuses specifically on analyzing the lunar surface. By making the system public, the two organizations are offering scientists a more efficient way to examine geological features and surface properties.

This development represents a notable shift in the tools available for space exploration. The AI was trained on more than 30 layers of data gathered by nine instruments across four NASA missions, creating a system capable of addressing intricate scientific questions. The training set includes roughly 2 million image tiles, which comprise over 1 million high-resolution camera images and nearly 964,000 multispectral images. This vast collection of data allows the model to produce detailed insights that support ongoing exploration efforts.

The project is the result of a partnership between NASA, IBM Research, and several academic institutions. To ensure easy access, the system is hosted on Hugging Face, where researchers can experiment with it for their own lunar studies. Its design is flexible, allowing it to be adapted for tasks such as mapping craters, identifying volcanic formations, and assessing the stability of ice deposits at the lunar poles. Because it requires very little labeled data to function, the model can be used across a wide variety of research goals.

For the scientific community, the NASA-IBM Lunar Foundation Model offers a way to simplify data processing, letting researchers spend more time on interpretation and analysis. By automating the identification and measurement of features like craters, the tool helps scientists piece together the Moon’s geological history and its place in the solar system more quickly. In testing, the model has consistently met or outperformed existing baseline systems across various applications.

Looking ahead, the model marks a new phase in how artificial intelligence is used within space science. It falls under a broader effort by NASA's Office of the Chief Science Data Officer to incorporate AI into scientific discovery. As the model is refined, it may help uncover new information about the Moon and its resources. This work between NASA and IBM establishes a framework for future AI applications, suggesting a more streamlined path for scientific inquiry in space.