Challenge
For over seventy years, NASA’s telescopes, orbiters, landers, and satellites have returned more data about our planet and the cosmos than any institution in history. Only a fraction has been truly examined. These holdings now exceed a scale and complexity that traditional computational analysis methods cannot fully exploit; scientists have found planets, transient events, and physical phenomena in data previously thought fully analyzed. By processing and integrating more than 150 petabytes of heterogeneous data, scientists will connect observations, simulations, and models across disciplines, time scales, and physical regimes in ways that accelerate discovery.
AI Solution
Recent advancements in artificial intelligence, including agentic and generative approaches, physics-informed AI, multimodal AI, specialized AI models, and scientific AI workflows present an opportunity to accelerate analysis and discovery across NASA’s research activities and science mission data. These tools will fuse observations, experimental data, simulations, and models from different instruments, missions, and scientific and physical domains, identify patterns and relationships across heterogeneous data sets, and improve prediction, interpretation, and hypothesis generation, particularly in areas such as space weather prediction, earth observation, and astrophysics.
Justification
NASA is uniquely positioned to contribute more than 150 petabytes of aeronautics research and science data, mission-driven research problems, and deep experience in data processing, data management, and scientific and engineering collaboration across aeronautics, astrophysics, heliophysics, planetary science, space weather research, Earth science, and biological sciences. The agency also brings fundamental expertise in developing foundation models based on scientific and engineering data. DOE is uniquely positioned to contribute complementary strengths in scientific data, high-performance computing, specialized AI models, scientific AI workflows, and methods emerging from existing DOE challenge areas. Together, these complementary assets can support AI-enabled scientific capabilities for integrating observations, simulations, and models to advance analysis, prediction, and discovery across NASA’s research and science mission datasets.
National Impact
Turning America’s mission archives into engines of discovery multiplies the return on every dollar the United States has ever spent in space. AI-enabled workflows which accelerate scientific and engineering discovery will improve the nation’s ability to understand Earth and space systems and expand the value of one of the Federal government’s most important science and engineering data resources. The resulting advances in AI-enabled scientific analysis will also support new approaches to prediction, modeling, and discovery across other data-intensive research domains.
This was last updated .