NASA launches Prithvi Geospatial AI model into orbit
Updated
Updated · Zamin · May 2
NASA launches Prithvi Geospatial AI model into orbit
11 articles · Updated · Zamin · May 2
Researchers at the University of Adelaide and SmartSat CRC tested NASA and IBM's model on the Kanyini satellite and an ISS module, detecting clouds and floods in space.
Trained on 13 years of Landsat and Sentinel-2 observations, the open-source system aims to cut data transmissions to Earth and speed disaster monitoring and agricultural assessment.
NASA says open-sourcing such models should accelerate science, and after its 2025 Surya solar model plans similar AI systems for astrophysics, planetary science and biology.
With AI now processing data in orbit, how soon can we directly ask satellites about unfolding disasters on Earth?
NASA and Europe are open-sourcing space AI. Who governs this powerful technology when it operates beyond national borders?
As AI proves its worth in orbit, can it overcome the immense physical challenges of heat and radiation for more complex missions?
Deploying Prithvi: NASA and IBM’s Breakthrough AI Model Transforming Satellite Disaster Response
Overview
Between 2024 and 2026, NASA and IBM achieved a historic milestone by deploying the Prithvi Geospatial AI model in orbit, running it simultaneously on the Kanyini satellite and the ISS. Designed to handle multiple geospatial tasks like flood detection and crop monitoring, Prithvi addressed the urgent need for faster Earth observation amid rising extreme weather events. Engineers overcame bandwidth limits with a compact update module, enabling continuous improvements. Building on this success, the advanced Prithvi-EO-2.0 model emerged in late 2024, dominating benchmarks and supporting real-world disaster responses, ecosystem monitoring, and urban heat mapping. This breakthrough paved the way for NASA’s broader AI strategy, including the Surya heliophysics model and open-source innovations.