In July 2026, the Lehigh University research team of Javad Khazaei, Associate Professor, Electrical and Computer Engineering; Shalinee Kishore, Iacocca Chair Professor, Electrical and Computer Engineering and Director of the Center for Advancing Community Electrification Solutions; and Farrah Moazeni, Assistant Professor, Civil and Environmental Engineering was awarded $518,000 for a three-year research project by the National Science Foundation’s Division of Electrical, Communications, and Cyber Systems (NSF ECCS). The proposal, titled “Stability at the Edge: Cyber-Physical Risks and Control of Hyperscale Data Centers in Future Inverter-Dominated Grids” falls under the the Energy, Power, Control, and Learning (EPCL) opportunity, which is designed to support the design and analysis of complex engineered systems. The funding represents the first extramural award for the ACES university research center. Khazaei, Kishore, and Moazeni are three of ACES’s founding members.
The stimulus for the proposal arose from a symposium hosted by Lehigh University in October 2025. Titled “Powering What’s Ahead: Energy and Water Challenges of AI Data Centers,” the event brought leaders from academia, government, and industry together on Lehigh’s Mountaintop campus to discuss issues surrounding resources for the growing data center infrastructure in the United States and specifically in Pennsylvania where many data centers, from hyperscale to small, are planned. (Amazon had recently announced plans to build in the Commonwealth when the symposium was held.) Khazaei says the symposium gave him and the ACES team the opportunity to listen closely to “the problems stakeholders face from the reliability and security of data centers to the grid operation perspective.” Those insights, as well as Lehigh’s strong advocacy of AI technologies, motivated this and other proposals.
In the past few years electricity use has increased sharply, driven by increased demands for air conditioning, data centers, and industrial needs and experts predict the trend will continue. One study shows data center energy consumption alone will grow 300% over the next 10 years. The production and distribution of all that energy will take a toll on existing infrastructure if innovations are not found. This is the focus area for the ACES team on this award.
Khazaei says that the energy needs of hyperscale data centers, which can cover millions of square feet, are huge. At times, such a facility can create a big surge in demand. The traditional, power-plant-based grid responds to a sudden increase in power demand fairly slowly, over a couple of seconds. This slow response results from the technology that’s used in the grid, and without a change to a new technology, this “demand response” will remain slow. The new, faster technology, or inverter-based resources (IBR), are based on power electronics and focused on the generation, or production, side. Hyperscale data centers can ramp up load demand very quickly, and currently, researchers and power operators don’t have an understanding of how these fast loads will impact the stability of the system. The main idea of this research, Khazaei says, “is to integrate the large-scale data centers into the power system and study how these super-fast data centers can impact the stability and resilience of the grid.” That’s what the name of the proposal, “stability at the edge,” refers to: “We’re integrating so many data centers into the grid without analyzing the impact, and we don’t know if this power system can handle it,” says Khazaei.
The first step, and Khazaei’s primary task for this research, is to create a full model of a data center, including the AI load, the uninterruptible power supplies, the cooling system, and the power system. A detailed model of the whole system will allow him to analyze what can go wrong and find out where instabilities arise. That’s where he’ll begin to look at failures. He says, “We want to see how data center power consumption could cause a blackout in a specific region in the power system. Or how could it trip, for example, transmission lines? How could it trip different loads, generators, and so on? What would be the impact of failures on generation on the power system of the future with larger-scale data center integration?” This second stage of the project, the “demand response” angle, will be handled by Kishore. This phase will consider what approaches can be taken by data centers to modulate their demand in order to not just reduce their demand on the grid but to opportunistically provide resources to help improve grid stability. “This will leverage work from the project’s first phase to understand what the vulnerabilities are and how demand flexibility can help alleviate them,” says Kishore.
The final phase, analyzing the impact of cyber attacks on power systems, falls under the purview of Moazeni. She will be looking at “load altering attacks,” or attempts to create a false load in order to induce instability in the grid. Khazaei says Moazeni “will be taking the lead on designing advanced controllers that would mitigate the impact of load altering attacks and stability issues.” Moazeni points out that national security is a concern because of “the scale and the speed of these energy-consuming components because if they are attacked, the impact is going to be way more devastating than somebody attacking an EV station, for example.” She adds, “We want to enhance the reliability of the grid in the presence of these data centers, making sure that the grid is still secure and safe.”
All of this falls squarely into the mission of ACES, particularly its focus on helping communities thrive via improved electrification systems. Data centers are part of a community’s electrification system, and like all users of the system, they require a reliable and stable grid. As Moazeni says, “We are enhancing the stability of the grid in the presence of these data centers, and making sure that we are still secure and safe.”
As society becomes more familiar with AI and uses it more frequently, data centers and energy consumption will rise. The positives for AI use, as Khazaei says, “are enormous” and “inevitable.” As he says, “We can't just say we don't want AI anymore. It's helping us. It's shifting how we think. It's shifting what the future is going to look like.” One of the first things AI will change, then, is the power systems that supply the data centers and the servers and other accessories that keep them running and keep them secure. For researchers at ACES, says Khazaei, “We have to see how we can enable this technology to go on.” And this research will be one step in developing and refining the systems that power not only AI but also the vast array of advanced electronics of the future.
“Data centers are something we need,” says Khazaei. “AI is useful for the future generation. We want to explore how AI could impact and shift the power system operation from being very slow to being able to respond very fast.”