NASA has partnered with IBM Research and several academic institutions to build an AI model trained on nearly 20 years of lunar orbiter data. The NASA-IBM Lunar Foundation Model is designed to speed up analysis of the moon's surface, helping researchers identify landing sites and locate water ice deposits faster than manual review allows.
How the model processes lunar data
The AI system searches for specific surface characteristics across massive volumes of imagery and scientific measurements. A researcher looking for craters of a certain depth with particular shadow patterns can get results in minutes or hours. That same search would take much longer using traditional methods. "We're going to be able to find things so much faster," said Zachary Aubert, founder and CEO of The Launch Pad Network.
The model was built using data from lunar orbiters collected over two decades. NASA's associate administrator for the Science Mission Directorate, Nikki Fox, said the agency sees AI as a way to extract new findings from its archives. "We also have to make data easier for scientists to explore and use. The NASA-IBM Lunar Foundation Model shows what's possible when we bring AI to NASA's petabytes of scientific data. That's a real opportunity we see with AI: turning large-scale data into new discoveries."
Scientists still drive the analysis
The technology handles the labor-intensive sorting work. It does not replace researchers. "It does the big comb through. It does the ugly work looking through," Aubert said. Scientists then examine the results and decide where to focus future missions. Aubert added, "You start to actually use your scientists for a much more detailed analysis."
Experts caution that AI output still requires human judgment. "It's going to help us find things faster, but it's also AI. We're still learning AI models," Aubert said. The agency has not indicated whether the initiative will affect staffing.
Public access and broader context
The Lunar Foundation Model is publicly available for researchers, educators, and the public to use. The release comes as both the United States and China push forward with lunar exploration programs, making rapid analysis of scientific data a higher priority for identifying exploration targets.
Why this matters for science and research professionals
This project demonstrates a practical pattern for AI for Science & Research: training models on large, domain-specific archives to accelerate Data Analysis without removing scientists from the decision loop. The approach converts time-intensive manual search into a task that runs in minutes, freeing researchers for interpretation and hypothesis testing. For teams managing large scientific datasets, the model offers a reference architecture for building similar tools in other fields.
Your membership also unlocks:
