Helping Power and Protect the Next Generation of AI Data Centers in North Carolina
By Michelle Fiscus, Senior VP & Chief Communications Officer
As artificial intelligence transforms nearly every industry, the demand for computing power is growing at an unprecedented pace.
For Dr. Tiefu Zhao, professor of electrical and computer engineering at the University of North Carolina at Charlotte, that rapid growth presents both a technical challenge and a strategic opportunity for North Carolina.
Zhao is leading an NCInnovation-funded project to develop technology that helps power and protect the next generation of AI data centers. His team's solid-state protection system is designed to make emerging 800-volt direct current (DC) data centers safer, more reliable and more efficient, while helping position North Carolina to become a supplier of the technologies behind the AI economy—not simply a place where data centers are built.
Every search, streamed movie, cloud application and AI prompt relies on data centers processing enormous amounts of information. As those facilities become more powerful, they also require dramatically more electricity. Technology companies are redesigning how data centers distribute power, shifting toward 800-volt direct current systems capable of supporting increasingly demanding AI workloads.
That transition creates a challenge current protection systems were never designed to solve.
"The biggest risk is that DC faults rise extremely fast and can create sustained electrical arcs," Zhao said. "At 800 volts, a fault can escalate in microseconds."
Unlike alternating current, direct current does not naturally pass through zero, making faults significantly harder to interrupt. Left unchecked, those faults can damage equipment, create fire hazards, and bring critical systems offline.
For Zhao, this isn't a new problem. It's one he has spent nearly a decade preparing to solve.
Before joining UNC Charlotte, Zhao led power electronics research and development at Eaton Corporation. When he transitioned to academia, he focused his research on technologies that could solve real industry challenges—not someday, but within a realistic commercialization window.
"We look at technology that can be commercialized in five to ten years," Zhao said.
Over the past ten years, supported by research funding from the U.S. Department of Energy, the U.S. Department of Transportation, and other partners, Zhao's team has advanced its technology from an early research concept to a working prototype.
Their solid-state protection system responds approximately 1,000 times faster than conventional mechanical circuit breakers while incorporating AI-based arc-fault detection to identify dangerous conditions before they spread. The design also uses commercially available semiconductor components, reducing cost while improving efficiency.
"Our advantage is not just speed," Zhao said. "It is the combination of speed, efficiency, cost, and intelligence."
The engineering, however, was only part of the journey.
“University researchers are very strong at advancing technology,” Zhao said. “The next challenge is building the right commercialization path, working with customers, industry partners and investors to turn that technology into a product.”
Like many university innovations, Zhao's technology had reached the point where another research grant alone was no longer what it needed most.
It needed a path to market.
Previous funding helped prove the technology worked.
NCInnovation is helping prove it can become a product.
Through NCInnovation, Zhao's team is building the commercialization strategy needed to move beyond the laboratory. Commercialization experts, industry advisors and an Entrepreneur-in-Residence are helping develop the business case, identify customers and prepare the technology for pilot deployments, certification and future partnerships.
“In the past, our business and commercialization team was relatively limited,” Zhao said. “Now with NCInnovation, we have a much stronger team to help us think about customers, market strategy and the path to scale.”
He believes those commercialization resources may be every bit as important as the technical work itself.
"We have done almost ten years of research on this technology," Zhao said. "At this stage, the business cases and value proposition are just as important as the technical validation.”
The timing matters.
AI infrastructure is expanding rapidly, and Zhao believes the opportunity to move university research into the marketplace will not remain open indefinitely.
"Without NCInnovation funding, the project would likely remain a strong university technology," he said. "Its transition to commercialization would be significantly delayed."
North Carolina is already attracting significant investment in data centers, electrification, and energy infrastructure. Zhao believes the state has the opportunity to do more than host that growth.
"This project helps the state move beyond simply hosting that growth," he said. "It creates the opportunity for North Carolina to supply one of the enabling technologies behind that growth."
That vision is one reason Zhao chose UNC Charlotte.
"When I transitioned from industry to academia, I was looking for a place that had a very specific focus on energy and power electronics," he said. "EPIC was one of the key reasons I came to UNC Charlotte."
Located on the UNC Charlotte campus, the Energy Production and Infrastructure Center (EPIC) is a state-of-the-art energy research and education center at the forefront of power electronics and power systems research, workforce development, and industry collaboration.
Today, discussions are already underway with industry leaders as Zhao's team evaluates the best commercialization pathway, whether through licensing, co-development, or a startup. The destination remains open, but the goal is clear: move the technology from university research into real-world deployment.
For Zhao, success will not simply be measured by papers, patents, or academic impact.
It will be measured by whether the technology creates real industrial impact, and whether North Carolina becomes known not only for building AI data centers, but for building the technologies that power and protect them.
"We can be the provider of the technology that powers future AI data centers," Zhao said.