Oak Ridge National Laboratory (ORNL) researchers have developed a new high-efficiency power converter using gallium nitride (GaN) semiconductors, a breakthrough that could help meet the growing energy and infrastructure demands of artificial intelligence (AI) data centers while improving power management systems across the energy sector.
Power converters are essential electronic devices that regulate and transform electrical current and voltage, enabling efficient power flow between energy sources, equipment and end users. Traditionally, these systems rely on silicon semiconductors. However, GaN semiconductors offer significantly faster switching speeds and lower energy losses, making them increasingly attractive for next-generation energy applications.
The ORNL team incorporated GaN semiconductors supplied by ROHM Semiconductor into converter systems developed at the laboratory’s Grid Research Innovation and Development Center (GRID-C). According to the researchers, the new converters can switch 10 to 20 times faster than conventional silicon-based devices while reducing energy losses during operation.
The compact design of the converter is another major advantage. Smaller and lighter systems can lower transportation, installation and maintenance costs while enabling more flexible infrastructure deployment for large-scale facilities.
The technology is particularly relevant for AI data centers, which are rapidly emerging as major electricity consumers globally. ORNL researcher Prasad Kandula noted that modern AI servers may require multiple power converters per unit, with enterprise-scale facilities deploying hundreds or even thousands of servers. In such environments, improvements in size, efficiency and thermal performance can significantly reduce operational costs and energy consumption.
Beyond AI infrastructure, GaN-based power electronics are also gaining traction in renewable energy systems, battery storage, EV charging infrastructure and smart grid applications due to their higher efficiency and reduced heat generation.
The development reflects a broader industry shift toward advanced semiconductor technologies that can support increasingly power-intensive digital and clean energy ecosystems.
According to the U.S. Department of Energy’s Oak Ridge National Laboratory, the project was conducted at GRID-C, a specialized research platform focused on grid systems integration, energy storage and advanced power technologies.
