WASHINGTON, September 10 — IBM and NASA on Thursday released the NASA-IBM Lunar Foundation Model, one of the first publicly available foundation models built for scientific exploration of the Moon, and with it the first open-source lunar dataset of its kind. The model is trained on an extensive lunar observation dataset curated by researchers at both organizations. According to the announcement, it exceeds widely used methods by up to 23% in identifying key geographic features on the lunar surface, including potential ice deposits, craters, and volcanic formations.
Sensors and instruments have observed the Moon for decades, generating petabytes of data. But scientists have had to either manually sift through maps and images or rely on low-resolution, task-specific machine learning models that can be computationally intensive and lack the scientific accuracy needed to identify and analyze geographic features.
Think of it like a library with millions of unlabeled photographs. The pictures are all there, but finding the handful that matter means flipping through the whole collection by hand. The new model is meant to be the index.
Kevin Murphy, chief science data officer and acting chief data and AI officer at NASA Headquarters in Washington, framed the release as a data-access problem as much as a technology one. “NASA has spent decades building an extraordinary scientific record of the Moon, but collecting data is only part of the job,” he said.
“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.”
What ships alongside the model
IBM and NASA scientists also built a unified, machine-learning-ready dataset that aggregates over 30 spatially-aligned layers from nine instruments across four missions. It combines tens of thousands of images and maps showing unique geophysical properties of the lunar surface, drawn from NASA’s Lunar Reconnaissance Orbiter and other sources. The model is designed to help researchers accelerate scientific progress by identifying hidden relationships between many different types and resolutions of lunar data.
In plain terms, it connects observations across instruments that are difficult to see in isolation. Juan Bernabe-Moreno, Director of IBM Research Europe, UK and Ireland, said the goal is to give the global research community something to build on.
“Uncovering the mysteries of the Moon requires an ability to learn from an extraordinary volume of scientific data,” he said. “The NASA-IBM Lunar Foundation Model gives scientists a foundation to explore the Moon at scale, connecting observations across instruments, revealing patterns that are difficult to see in isolation, and providing an open platform the global research community can build on.” Both the model and the dataset are now available as open-source resources.
The practical upshot
The Moon’s topography is ever-changing, with craters, distributed ice, and volcanic activity shaping its surface over time. The new foundation model is expected to help researchers investigate multiple lunar phenomena and support a sustained return to the Moon. Watch for how quickly outside research groups put the open dataset to work, and whether the model’s feature-detection edge holds up on lunar terrain it has not seen before.
