This project is inventing ways to train and run giant science AI models far faster and with much less energy—using smart compression on supercomputers—so they’re cheaper to build, easier to use, and don’t crowd out other critical research.

Researchers are modeling coolant flow and heat transfer in the thin, curved channels of research reactors—providing safer design benchmarks and helping convert reactors from highly enriched to safer low-enriched uranium fuel.

Fermilab’s Short-Baseline Neutrino program is using multiple cutting-edge detectors and supercomputer-powered AI to sift petabytes of data on shape-shifting neutrinos—settling puzzling anomalies that could hint at new particles and guiding next-generation experiments.

This project connects the upgraded Advanced Photon Source’s brighter x-ray experiments to supercomputers—processing huge datasets in near real time with shared workflows—to speed breakthroughs across materials, biology, energy, and more.

Researchers are running next-generation plasma simulations to see whether Z-pinch devices—which squeeze super-hot gas with magnetic fields—can be stabilized, offering clearer guidance toward practical, reliable fusion energy.

Sandia and NREL are simulating next-gen aircraft engines running on synthetic fuels, predicting soot and flame stability to cut certification risk and speed cleaner, home-grown jet fuels to market.

This project tests a new glass-like boron carbide “shell” for fusion fuel capsules to keep the fuel cleaner during implosion and unlock much higher energy output.

EMERGE uses exascale simulations and AI to model how diseases spread through people and the environment, delivering rapid, uncertainty-aware forecasts and testing intervention strategies so public-health officials can make better real-time decisions.

This project uses supercomputer simulations to study and tame tricky airflow inside the curving engine inlets of next-generation blended-wing aircraft—testing smart flow-control techniques to keep air smooth, protect engines, and boost fuel efficiency.

This project is training AI coding assistants built for supercomputing—able to read huge codebases, generate and optimize parallel programs across diverse hardware, and explain their choices—to speed up and strengthen scientific software development.

Researchers are modeling how tiny twists and layerings of ultra-thin materials create new electronic behaviors—building an open “moiré kaleidoscope” database to speed the discovery of next-generation materials like superconductors.

Using supercomputers, researchers will simulate how hot helium mixes with air during rare cooling emergencies in advanced reactors, producing data that helps engineers design safer systems and prevent oxygen-related risks.

This project is creating AI models that know their limits, explain their reasoning and uncertainty, and safely collaborate with people and tools—so we can trust them in high-stakes science and industry.

This project uses remote supercomputers to analyze DIII-D fusion experiments in near real time—feeding results back to the control room to guide decisions, protect equipment, and accelerate progress toward practical fusion energy.

Using DOE supercomputers, researchers are training powerful AI models that learn from text, images, and energy data across many institutions—without sharing sensitive data—to accelerate discovery, strengthen the power grid, and enable secure scientific collaboration.

Using one of the world’s fastest supercomputers, researchers will simulate how fuel burns and hot gases flow through a jet engine’s combustor and turbine together—revealing physics that can boost efficiency, cut emissions, and enable tougher, longer-lasting designs.

This project uses powerful supercomputers to rapidly screen and simulate new battery liquids, helping scientists design safer, longer-lasting, high-energy batteries for cars and devices by understanding how lithium moves through next-generation electrolytes.

This project uses powerful supercomputers to simulate how bubbles—and soap-like additives that change their behavior—move and mix in fast, churning flows, yielding insights that can make reactors, heat exchangers, and cooling systems safer and more energy-efficient.

This project develops a resilient, human-like AI that learns and reasons on the fly—using supercomputers and diverse space and fusion data—to help satellites safely dodge debris and keep fusion reactors running smoothly, strengthening national security and accelerating clean energy.

TAE Technologies is using advanced supercomputer simulations to test and refine its field-reversed-configuration fusion approach—building on experiments that show wider stable operating ranges and better energy confinement at higher temperatures—to determine if these gains hold at power-plant conditions, reducing risk and bringing carbon-free fusion electricity closer to reality.