Better Brains: Meet the Engineer Helping to Power Complex New AI Data Centers

As artificial intelligence scales, so does the challenge of powering it. New data centers demand more advanced power and control systems that reliably deliver massive quantities of energy. Researchers are developing flexible, efficient infrastructure designed for the next generation of computing.

The AI data centers breaking ground around the world are so massive that they're described as "hyperscale" - spanning hundreds of acres and incorporating hundreds of thousands of chips. Unlike typical cloud data centers, which run many independent processes, new AI data centers harness colossal numbers of servers and computer cores. They must work in precise coordination to run large-scale AI models.

Because of their demands, supplying power to these data centers requires new strategies. At GE Vernova, one innovative approach employs solid-state transformers. In simple terms, transformers convert electricity from the grid into a usable form. Traditional ones do this passively with large coils of metal, while solid-state transformers use fast power semiconductor switches to do the same job with more precision and flexibility. These "SSTs" employ advanced algorithms to monitor, control, and coordinate power flow - even when things go wrong.

That's where Mike Englert comes in. Englert, a principal engineer at GE Vernova Advanced Research in Niskayuna, New York, works on the Embedded Systems team designing the "brains," or controllers, of these SSTs. His work focuses on advancing real-time deterministic, low-latency control.

Rather than relying on traditional centralized control, Englert's research explores embedding intelligence directly into the communication fabric. This approach reduces delays from networking, software, and operating systems, enabling thousands of distributed devices to sense and act within microseconds. It is designed to support faster, more predictable system behavior while laying the foundation for an adaptable control platform across a wide range of energy systems.

Englert and his team at GE Vernova Advanced Research in Niskayuna. Images credit: GE Vernova

"We're trying to power warehouses and warehouses of compute," Englert says, "from AI training and inference to cloud services and advanced industrial workloads. That means coordinating tens of thousands of power conversion elements so they operate in precise lockstep. A huge number of controls - thousands of devices - need to be coordinated properly to transfer power with the appropriate flexibility and reliability to protect the data center server racks."

Designing the "Brains" of AI Data Center Power Systems

How do Englert and his team improve this synchronization? One way to think about the challenge is conducting an orchestra rather than playing a single instrument. Each solid-state transformer must continuously respond to changing conditions but also remain synchronized with every other unit in the system. Sensors provide real-time measurements, such as voltage, current, and temperature. The control system uses this data to adjust the precise timing of electronic switching events. When everything is properly coordinated, the system settles into a stable, repeating pattern that delivers consistent and efficient power.

"If you can make the 'brain' smarter and faster, you can have power converters that are smaller and more power dense," Englert says. "This means the components can be smaller and costs can come down."

Dreaming Big

For Englert, working on this technology has been a lifelong dream. He grew up in the Albany area, near Niskayuna, and knew his career goal from an early age. (His mother has proof: a crayon picture of a roller coaster that Englert drew in second grade, next to the word "engineer.") He joined GE Research, as it was called then, in 2016. Being close to his hometown allows him to see his family often - he even still helps out at his older twin brothers' pizzerias - and to see the lasting impact of his work.

"If you want to be the best at science in this area, this is where you go," Englert says. He and his colleagues refer to the research center as "Candyland," a place where they can drop into the offices of co-workers with advanced degrees in thermal management, mechanical engineering, and electrical engineering. "All the projects and teams are interdisciplinary, and you have all sorts of experts at your fingertips," he explains. "It's almost like a university campus. Put a bunch of passionate, smart people together and you make magic."

This environment motivates Englert to push technology beyond its current limits. He wants to help position GE Vernova as a leader in next-generation control architectures by developing what he describes as a "future-proof" control platform - one that can scale across applications and adapt to emerging technologies. By unifying computation, communication, and control into a cohesive system, Englert envisions a platform that not only improves today's power converters but also goes further. He aims to enable new capabilities in areas like energy storage, renewable integration, and high-density power systems for AI infrastructure.

"Innovation should be cross-cutting, and bringing people along who are interested in the same goal is exciting," he says. "Research is a team sport."

As AI continues to scale, so will the demand for smarter, faster, and more resilient energy systems. In Niskayuna, Englert and his team are helping make that future possible - one system, one breakthrough, and one team at a time.