IBM and NASA launch open-source AI model for lunar exploration

IBM and NASA have unveiled a new open-source AI foundation model designed to accelerate lunar research. Available on Hugging Face, the model allows scientists to analyze decades of lunar data to identify geological features, such as ice deposits and craters, which are critical for establishing a sustained human presence on the Moon. By moving away from task-specific models, researchers can now utilize a unified system to study petabytes of data from NASA’s Lunar Reconnaissance Orbiter and GRAIL missions. This release includes a massive, multimodal dataset, marking the first time such a comprehensive cache has been made available for machine learning. The initiative aims to improve the accuracy of surface analysis, helping future astronauts navigate safely, locate essential resources, and better understand the Moon's volcanic and thermal history. This collaboration underscores a long-standing partnership between IBM and NASA, providing the global research community with tools to advance space exploration.