The U.S. Department of Energy's Genesis Mission is building the American Science Cloud to accelerate AI-driven scientific discovery. To support this national effort, Fermilab's Fermi Data Platform is providing the secure, large-scale storage infrastructure required to make advanced AI research possible across the country.
Building the data backbone
The Genesis Mission requires more than supercomputers. Researchers need a secure, high-performance infrastructure to store and access massive datasets efficiently. Fermilab was selected as a key partner to provide petabytes of storage, advanced data-access tools and institutional expertise.
Fermilab has decades of experience managing immense datasets for projects like the CMS experiment at CERN's Large Hadron Collider and the Short Baseline Neutrino program. The lab is also preparing its infrastructure for the upcoming Deep Underground Neutrino Experiment.
"At Fermilab, we orchestrate thousands of disks to provide petabytes of storage space, and we make sure researchers can access their data quickly and securely," said Oliver Gutsche, lead of the Fermi Data Platform project. This centralized approach prevents bottlenecks when multiple research teams query the same datasets.
Making data AI-ready
For AI systems to function at scale, they need data that is accessible, well-organized and prepared for machine learning. Raw scientific data from instruments often lacks the structure and metadata that models require. The Fermi Data Platform bridges this gap by storing datasets from Genesis Mission projects and presenting them for model training and inference.
This infrastructure allows scientists to tap into a system where they can describe a research need and have AI-driven tools access national laboratory resources. For professionals looking to adapt to these shifts, understanding the AI Learning Path for Research Scientists can provide practical frameworks for integrating these tools. The goal is to automate intermediate steps like literature searches and preliminary simulations.
"Give me the 10 most promising materials for batteries - the system does a literature search, runs some simulations to verify, narrows down that list, and presents it as an answer for further research," said Gutsche. "That is the kind of workflow the Genesis Mission is designed to enable."
The goal is to reduce the time between asking a scientific question and getting a meaningful answer. Humans still ask the questions and evaluate the results, but the system allows scientists to work faster and focus on higher-level insights.
"Data is the common denominator behind major scientific endeavors, and AI is fundamentally data-driven," said Chin Guok, partner integration level 1 lead for the American Science Cloud. "To train and run AI models, you need large volumes of data. Fermi Data Platform can support AI training and inference on large scientific datasets." This partnership allows researchers to use DOE resources more directly, laying a foundation for accelerated scientific discovery.
Why this matters for science and research professionals
The integration of large-scale data platforms with AI services will change how research workflows are executed. Scientists in fields like high-energy physics, materials science, and fusion-energy research will increasingly rely on automated systems to filter raw data and surface targeted insights. Understanding how to structure and query data for model training will become a core competency for researchers using AI for Science & Research.
Your membership also unlocks:
