Arizona State University to study nuclear power for AI data centers in DOE program
Story Highlights
- Arizona State University and DCX will study microreactor energy for AI
- AI data centers may consume one-eighth of U.S. electricity soon
- The Energy Department selected five research teams for MARVEL program
The U.S. Department of Energy has selected Arizona State University and DCX USA, LLC, as key research partners for its Microreactor Application Research Validation and Evaluation (MARVEL) program, an innovative national effort led by Idaho National Laboratory to establish novel applications for advanced microreactor technologies.
A joint ASU–DCX research team will study whether a small modular reactor can provide dependable energy for the growing demands of artificial intelligence and high-performance computing, strengthening U.S. competitiveness in AI.
This landmark project will generate key performance data on microreactor reliability, load-following capability and stability under artificial intelligence workloads.
“At ASU, energy research and education are focused on exploring and advancing a diverse portfolio of resilient and viable energy solutions,” said Sally C. Morton, executive vice president of ASU Knowledge Enterprise. “Now, more than ever, we need collaborative and vigorous innovation to solve the challenges of an evolving energy future.”
AI data centers are among the most energy-intensive computational facilities in the world, and meeting future AI power requirements is widely recognized as a national priority. They are projected to consume up to one-eighth of the total electricity in the U.S. in the next few years — equal to the current power use of California and Texas combined.
Scaling the nation’s energy supply to meet the growing demands of AI — while ensuring energy security, reliability and affordability — will be essential to U.S. security and leadership in AI.
“Data centers are the backbone of our digital future, and the energy they require is growing exponentially,” said George Slessman, founder of DCX. “Through this partnership with ASU and INL, we’re demonstrating that nuclear-powered, AI-optimized infrastructure is not only feasible — it’s essential. This work lays the foundation for sovereign, resilient and infinitely scalable AI capacity built here in the United States.”
The ASU-DCX research team will conduct an in-depth evaluation of microreactor-grid interactions, AI load modeling and thermal-to-electric system optimization. The study’s findings will help determine how small, transportable reactors can directly power modular data centers — an approach that could reduce dependence on traditional electrical grids, enable deployment of AI computing in remote or infrastructure-limited environments, and provide a continuous, carbon-free power source.