Oak Ridge sets up unit to tackle AI datacenter energy demand

theregister.com · Feb 27, 2026 · Collected from GDELT

Full Article

Oak Ridge National Laboratory (ORNL) is hoping to turn its technical expertise to the problem of growing electricity demand from AI datacenters. The US federal research center in Tennessee is launching the Next Generation Data Centers Institute (NGDCI) in support of national energy and AI priorities, though any lessons learned will likely apply globally as well. As an internal "institute within an institute," NGDCI aims to bring together the laboratory's facilities and expertise spanning energy, high-performance computing, cybersecurity, and grid technology. The ultimate goal is to ensure that America's rapidly growing AI infrastructure remains secure, efficient, and reliable, the lab says. "Artificial intelligence is transforming every part of our society, but its energy appetite is unlike anything we've seen before," ORNL Director Stephen Streiffer stated. "The electricity required to power AI datacenters is expected to double or triple in the coming decade, straining infrastructure that is already under pressure. ORNL is uniquely positioned to meet this challenge." Efforts will cover six research areas: thermal management, power system architecture, grid integration, security, integrated systems modeling, and operational load management. Thermal management will focus on the next generation of cooling tech, as ORNL claims current systems may account for 40 to 60 percent of a datacenter's total energy consumption. Research into power system architecture will look at how energy flows from source to server, bringing in ideas such as direct current and new power electronics to reduce losses. Security considerations will look at embedding cyber-informed engineering (an actual thing, apparently) into the infrastructure of datacenters, along with quantum-safe communications. Integrated systems modeling aims to look beyond the walls of the datacenter to anticipate how AI infrastructure may impact energy, jobs, materials, and US competitiveness through the 2030s and beyond. ORNL says that with the ballooning energy requirements of giant data campuses, the US grid cannot absorb the projected load growth without new approaches to planning and operation. But addressing the problem with intelligent integration that links power, cooling, thermal management, workload scheduling, and AI-enabled forecasting could turn it into an advantage. Amazon-backed X-Energy gets green light for mini reactor fuel production All aglow about DCs, investors launch $300M at microreactor startup HPC won't be an x86 monoculture forever – and it's starting to show The exascale offensive: America's race to rule AI HPC "We envision a future where datacenters are national assets: adaptive, efficient, and strengthening the nation's grid while fueling discovery and advancing America's leadership in AI," enthused Robert Wagner, ORNL associate laboratory director for energy science and technology. Notable names in the AI datacenter arena, including chipmakers Nvidia and AMD, welcomed the move. "For decades, national labs like ORNL have relied on close collaboration with the computing industry to advance high-performance computing, and the next generation of AI is redefining requirements at the intersection of compute, power, and the grid. NGDCI is designed to address those challenges at scale," commented Forrest Norrod, the head of AMD's Data Center Solutions Business. ORNL is also part of the "Genesis Mission," an initiative launched last year by the Trump administration as a national effort to develop and use AI for scientific discoveries, pooling supercomputers and other facilities from the numerous national labs with resources from industry and academia. As part of its contribution, the lab is preparing to deploy the Discovery and Lux supercomputers. Announced in October, Discovery is set to be a successor to Frontier, the world's first exascale system, while Lux is an AI cluster intended to advance machine learning. NGDCI will focus on the technologies required to operate these systems reliably while accelerating scientific breakthroughs, ORNL says. ®