University of Houston Engineers Discover Efficient Cooling Method for AI Data Centers
Engineers at the University of Houston have developed a new method to cool semiconductors three times more efficiently, which could drastically reduce energy and water consumption in AI data centers. This breakthrough aims to address the high resource demands of data centers, especially as Texas becomes a major hub. The research offers a sustainable approach to managing the heat generated by powerful AI computing systems.
University of Houston (UH) engineers, led by distinguished professor Hadi Ghasemi, have made a significant breakthrough in cooling technology that could dramatically reduce the energy and water consumption of AI data centers. Their research focuses on improving the efficiency of semiconductor cooling, which is critical given the intense heat generated by modern, increasingly powerful chips. The team, including doctoral candidates Amirmohammad Jahanbakhsh and Saber Badkoobeh Hezaveh, discovered that branched, tree-like microstructures in cooling systems can release heat three times more efficiently than conventional designs.
This innovation is particularly relevant for Texas, which is projected to become a major data center hub, facing growing public concern over the facilities' substantial electricity and water demands. Ghasemi emphasized the importance of a sustainable approach for the growth of AI, noting Houston's role as an energy supplier. The research, which began around 2020 and is ongoing, is partly funded by a $1 million grant from the U.S. Office of Naval Research and has been published in the International Journal of Heat and Mass Transfer.
Ramanan Krishnamoorti, UH Vice President of Energy and Innovation, highlighted the anticipated 30-50% increase in U.S. electrification over the next decade, with Texas alone potentially seeing a 35% rise. This increased demand is expected to lead to higher energy costs for consumers. The UH team is therefore focused not only on developing better technologies but also on making them cost-effective to mitigate impacts across socioeconomic classes. Ghasemi hopes the technology will eventually be implemented in data centers to ensure their sustainable expansion.