ORNL Launches the Next-Generation Data Centers Institute

Why It Matters

By addressing the energy and reliability challenges of AI data centers, NGDCI protects grid stability while enabling the United States to maintain leadership in AI and high‑performance computing.

Key Takeaways

• Data centers use over 4% of US electricity.
• AI workloads could push usage to 17% by 2030.
• NGDCI unites energy, HPC, cybersecurity expertise at ORNL.
• Industry partners include AMD, NVIDIA, Carrier, Chemours.
• Research focuses on cooling, power architecture, grid integration.

Pulse Analysis

The rapid expansion of artificial‑intelligence workloads is reshaping the energy landscape, with data centers already accounting for more than four percent of U.S. electricity consumption. Forecasts from the Electric Power Research Institute suggest that figure could climb to 17 percent by 2030, driven by the massive power draw of training large language models. In response, the federal government’s Genesis Mission is linking the nation’s most powerful supercomputers to the electric grid, aiming to double research productivity within a decade. This policy backdrop underscores the urgency of creating data‑center designs that consume less power while delivering higher performance.

NGDCI positions Oak Ridge as the hub for solving these challenges by leveraging its unique testbeds, such as the campus microgrid and the MEGA‑DC decision‑support platform. The institute’s six research thrusts—thermal management, power‑system architecture, grid integration, autonomous operations, security, and integrated systems modeling—target the full stack from silicon to facility. Collaborations with industry leaders like AMD, NVIDIA, Carrier and Chemours bring cutting‑edge cooling technologies, power‑aware architectures and cyber‑resilient designs into the lab, accelerating their path to commercial deployment and reducing the carbon footprint of future AI clusters.

Beyond immediate energy savings, NGDCI’s work promises strategic advantages for the United States. By turning data centers into grid assets that can provide demand response and ancillary services, the institute helps safeguard national energy security and supports the broader goal of maintaining global AI leadership. With projected global data‑center spending of $7 trillion by 2030—over 40 percent of which will flow to the United States—ORNL’s integrated approach could shape the next generation of resilient, low‑carbon AI infrastructure, driving economic growth and reinforcing the country’s competitive edge.