NASA and IBM have joined forces to develop a new artificial intelligence (AI) model that compiles and analyzes vast amounts of data about the Moon. This model, named the Lunar Foundation Model, is part of the Prithvi family of AI systems, which also includes models for Earth observation and weather analysis. According to a statement from September 10, 2026, the Lunar Foundation Model is now available as open source on platforms like Hugging Face and GitHub, making it accessible for researchers without the need for large additional datasets. The Lunar Foundation Model is a type of "foundation model," meaning it is trained on large amounts of raw, unlabeled data to learn the general structure and patterns of the Moon's surface. The model uses a visual encoder based on a Vision Transformer (ViT-B) paired with a decoder to process and interpret complex data. NASA reports that the AI successfully merged and aligned over thirty layers of diverse data from nine different instruments across four space missions, including NASA's Lunar Reconnaissance Orbiter (LRO). These data include high-resolution images, topographic maps, thermal readings, and shading, enabling the AI to uncover patterns that are not easily visible to humans or traditional algorithms. One of the model's standout features is its ability to detect permanently shadowed regions at the lunar poles, which are potential locations for water ice. This is critical for future lunar bases, as water is a key resource for sustaining human presence. The model has reduced error rates by 22 to 23% compared to existing systems in identifying these regions. It also outperforms conventional methods by 19% in mapping craters at a scale of about 100 meters, using only half the training data. This efficiency is especially valuable for the Artemis program, which aims to establish safe landing zones on the Moon. Additionally, the model has improved the detection of irregular volcanic structures by 3%, helping geologists better understand the Moon's volcanic history and thermal evolution. In a recent incident, a Falcon 9 rocket first stage from SpaceX crashed onto the Moon, creating a new impact crater. While the event was accidental, NASA noted that the Lunar Foundation Model could have identified the new crater in real time. This highlights the model's potential for monitoring lunar changes and supporting future space missions with more accurate and timely data.