IBM and NASA have launched an open-source artificial intelligence model designed to help scientists analyze decades of lunar observation data, identify ice deposits and map craters as the United States prepares for a sustained human presence on the Moon.
The NASA-IBM Lunar Foundation Model was trained on more than 30 layers of data collected by nine instruments across four NASA missions, including the Lunar Reconnaissance Orbiter. The model is part of the broader Prithvi family of open foundation models developed by IBM and NASA for applications including geospatial and weather analysis.
Read: NASA selects Relativity Space for privately funded Mars science mission
(June 18, 2026)
In benchmark testing, the AI model identified important lunar surface features with up to 23% greater accuracy than widely used existing methods, according to NASA and IBM.
The model could help researchers locate potential ice deposits in permanently shadowed regions of the Moon, map craters that could influence the selection of safe landing sites and identify volcanic formations. These tasks have traditionally required scientists to manually examine large quantities of maps and images or rely on lower-resolution machine-learning systems.
Why lunar ice matters
Scientists and space agencies are particularly interested in ice deposits because water could become an important resource for future lunar missions. Water can potentially be separated into hydrogen and oxygen, which could support life-support systems and the production of rocket fuel.
That makes mapping the Moon’s permanently shadowed regions an important part of planning for longer-term human exploration.
NASA’s Artemis program plans to return astronauts to the Moon in 2028 while testing technologies and systems intended to support a sustained lunar presence and eventual missions to Mars.
The new model is designed to make the growing volume of lunar data easier for researchers to analyze. By combining observations from multiple instruments and missions, the AI system could help scientists identify features that may otherwise take considerable time to locate manually.
Read: Indian American NASA astronaut Anil Menon makes first spaceflight (July 15, 2026)
The release also reflects a broader push by NASA and technology companies to use AI for space science, where increasingly large datasets are being generated by robotic missions.
For future lunar exploration, improved maps could help scientists and mission planners assess potential landing locations, understand the terrain and identify resources that could reduce the amount of material astronauts would need to transport from Earth.