By 2030, energy consumption of American AI data centers may increase fourfold 0 55

The American National Laboratory has announced the launch of the Next Generation Data Center Institute (NGDCI), a new large-scale initiative aimed at addressing the growing energy challenges associated with the operation of data centers for artificial intelligence. The new institute will bring together a wide range of expertise from the Oak Ridge National Laboratory (ORNL) in energy technologies, computing, networking science, and cybersecurity to create a safe, efficient, and reliable infrastructure for the future.

According to ORNL Director Steven Streyffer, artificial intelligence is transforming all areas of society, but its appetite for energy is unprecedented. It is expected that in the next decade, electricity consumption for powering AI data centers will double or triple, creating additional strain on an already overloaded infrastructure. According to an analysis conducted by the Electric Power Research Institute, data centers currently account for more than four percent of the total electricity consumption in the U.S., and by 2030, this figure could reach 17 percent. The primary reason for this growth is the development of artificial intelligence, as training and operating large models require enormous computational power.

The activities of NGDCI will focus on developing scientific and technical solutions in the areas of cooling systems, energy consumption management, and integration with energy grids, ensuring that future AI infrastructure does not jeopardize the reliability or security of energy supply. The institute's work will be based on an existing ORNL project modeling energy consumption growth related to data centers (MEGA-DC), which uses advanced tools to assess the economic and technical implications of infrastructure upgrades.

ORNL scientists envision a future where data centers become national assets capable of adapting in real-time and even enhancing the resilience of the energy system. To collaboratively develop solutions, NGDCI will bring together a wide range of industry partners. Major technology companies, including AMD, Carrier Energy, Chemours, and NVIDIA, have already expressed support for the initiative, emphasizing collaboration on energy-efficient architectures, advanced cooling technologies, and system security.

These collaborations aim to implement breakthrough developments using unique ORNL assets, such as microgrid testbeds, digital twin environments, and extensive energy capabilities. The launch of NGDCI coincides with ORNL's preparations for deploying two next-generation computing systems for AI — Discovery and Lux — further expanding the laboratory's role in the development of high-performance computing infrastructure. The institute will focus on the technologies necessary for reliable energy supply and operation of these systems at scale while also facilitating scientific discoveries. By combining energy research with advanced computing and national security priorities, NGDCI is set to play a key role in shaping the U.S. approach to building and operating future AI data centers, ensuring that they are not only powerful but also resilient and secure for the energy system.