Tianyu Wang (ECE) Advances Energy Efficiency in AI Data Centers

By Mia Knežević, CISE Staff Writer

Artificial intelligence already consumes a massive amount of energy — and the demand is only expected to grow. According to Energy Digital, data centers powering large-scale AI systems accounted for an estimated 1–1.5% of global electricity use in 2022. That number is projected to double by 2026.

Tianyu Wang, ENG assistant professor (ECE), and faculty affiliate of CISE and Hariri Institute is exploring new ways to build energy-efficient hardware to address this problem. By combining physics, optics, and computer science, Wang is developing technologies that could improve the efficiency of everything from large AI data centers to autonomous systems operating at the edge.

Wang’s research focuses on leveraging physical processes—such as optical phenomena—to perform neural network computations more efficiently. These analog systems, unlike traditional digital computers, hold promise for neuromorphic computing, a method well-suited for handling the demands of today’s large-scale AI models. By moving beyond conventional computing platforms, Wang aims to develop solutions that not only reduce energy consumption but also make advanced AI technologies more accessible and sustainable.

Learn more about his work in this CISE story.