A new Oak Ridge National Laboratory (ORNL) initiative aligns with the US federal strategy to strengthen AI infrastructure and grid reliability.

As electricity consumption from AI data centres accelerates across the United States, scientists at ORNL have established a new research institute focused on managing the energy, security, and operational pressures tied to that growth.

The laboratory this week announced the creation of the Next Generation Data Centers Institute (NGDCI), an internal programme designed to coordinate its work in high-performance computing, energy systems, cybersecurity, and grid modelling.

The move comes amid mounting concern that expanding AI infrastructure could strain power networks and complicate long-term energy planning.

ORNL Director Stephen Streiffer explained: “Artificial intelligence is transforming every part of our society, but its energy appetite is unlike anything we’ve seen before.

“The electricity required to power AI data centres 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.”

Rising electricity demand from AI data centres

AI data centres already account for more than 4% of total US electricity consumption. Analysts at the Electric Power Research Institute estimate that figure could rise to as much as 17% by 2030 if current expansion trends continue.

Much of the increase is driven by AI-specific computing tasks. Training large-scale machine learning models can require hundreds of megawatt-hours of electricity, and deployment at scale adds further load.

The North American Electric Reliability Corporation has warned that rapid demand growth from AI systems and industrial electrification may heighten risks to grid stability in several regions.

At the same time, investment in digital infrastructure continues to climb. Consultancy McKinsey & Company projects that global spending on data centres could reach $7 trillion by 2030, with the United States accounting for more than 40% of that total.

Federal Strategy and the Genesis Mission

The institute’s launch coincides with a broader federal effort led by the DOE under an initiative known as the Genesis Mission.

That programme aims to more closely link the nation’s advanced computing assets with the energy systems that power them, with the goal of significantly increasing research and development output over the next decade.

Researchers at Oak Ridge say the new institute will support those objectives by examining how AI infrastructure can be powered, cooled, and secured without undermining reliability.

The laboratory is preparing to deploy two next-generation AI supercomputers, Discovery and Lux, and the institute will focus in part on ensuring such systems operate efficiently within broader energy constraints.

From grid strain to grid support

One of the central challenges is that existing electricity networks were not designed to absorb the scale and concentration of load associated with modern data centres. Traditional planning approaches may prove insufficient if demand continues to rise at current rates.

The laboratory argues that smarter integration could change the equation. By coordinating power delivery, thermal management, workload scheduling, and AI-enabled forecasting, AI data centres could eventually help balance supply and demand rather than simply adding to the system’s stress.

To support this work, the institute will draw on modelling tools developed through Oak Ridge’s Modeling Energy Growth Associated with Data Centers (MEGA-DC) project.

The platform evaluates infrastructure costs, economic impacts, and upgrade pathways for utilities, state regulators, and developers assessing new data centre projects.

Six core research areas

The NGDCI will concentrate on six technical priorities:

• Advanced thermal management to reduce the substantial energy and water required for cooling, which can represent up to 60% of total facility consumption.
• New power system architectures aimed at cutting transmission losses and improving efficiency between the generation source and server rack.
• Grid integration strategies that explore how data centres might contribute to resilience rather than undermine it.
• Autonomous operations and load management systems to optimise computing workloads alongside energy availability.
• Cybersecurity measures, including quantum-resistant communications, to protect both physical and digital infrastructure.
• Integrated systems modelling to assess long-term impacts of AI infrastructure on energy markets, employment, materials demand, and US competitiveness.

As AI data centres continue to expand, policymakers and utilities face difficult trade-offs between economic growth and grid stability.

Oak Ridge’s new institute signals that national laboratories are positioning themselves at the centre of that debate, focusing on how to align advanced computing ambitions with the realities of the energy system that sustains them.