New computer chips could reduce energy use by artificial intelligence

The structure of ESO memristor: Each layer is a different material.

(JUNE 28, 2024) A researcher at Oregon State University has helped develop a new artificial intelligence chip that could improve energy efficiency by six times.

As the use of artificial intelligence (AI) increases, so does the amount of energy it requires. Projections show artificial intelligence will require half a percent of global energy use by 2027. In fact, it will use as much energy annually as the entire country of the Netherlands.

Sieun Chae is an assistant professor of electrical engineering and computer science. She is working to help reduce the technology’s electricity footprint. She is researching a new kind of chip that is based on an innovative material. It will allow for both computation and data storage. It imitates the way biological neural networks handle information storage and processing.

“With the emergence of AI, computers are forced to rapidly process and store large amounts of data,” Chae says. “AI chips are designed to compute tasks in memory, which minimizes the shuttling of data between memory and processor; thus, they can perform AI tasks more energy efficiently.”

The chips feature components called memristors. That is short for "memory resistors". Most memristors are made from a simple material system composed of two elements. However,  the ones in this study feature a new material system known as entropy-stabilized oxides, or ESOs. They have more than six elements, and that allows their memory capabilities to be adjusted in many ways.

Memristors are similar to biological neural networks. Neither has an external memory source. Thus, no energy is lost to moving data from the inside to the outside and back. By finding the ESO composition that works best for specific AI jobs, these chips can perform tasks with far less energy than a computer’s central processing unit.