As artificial intelligence expands, the data centers powering it face a growing challenge: how to meet rising computing demand without wasting energy.
Christina Delimitrou, a Greek associate professor at the Massachusetts Institute of Technology (MIT), is working on solutions that could help make these facilities more efficient. Her research explores how machine learning can improve the way large computing systems use their existing hardware.
Data centers support everything from cloud storage and video calls to the AI tools people increasingly use every day. Yet their growing energy needs are putting pressure on electricity grids and raising environmental concerns.
How AI Could Make Data Centers More Efficient
Delimitrou and her research group examine how servers, software, and computing resources work together. Their goal is to reduce inefficiencies and extract more computing power from existing equipment.
Research conducted during her graduate studies at Stanford University revealed that some large computing systems operated at only about 15% of their capacity. The finding highlighted how poorly utilized resources can contribute to inefficient energy use. Machine learning can help automate resource management and identify ways to distribute computing workloads more effectively. This could allow data centers to handle more applications with the same hardware, while maintaining performance and reliability.
Her group also develops tools that help programmers identify problems in cloud applications before they cause disruptions.
A Greek Researcher at the Forefront of Sustainable Computing
Delimitrou studied computer engineering at the National Technical University of Athens before pursuing graduate studies at Stanford. She later joined Cornell University as a faculty member and moved to MIT in 2022.
She is now an associate professor in MIT’s Department of Electrical Engineering and Computer Science and a member of its Computer Science and Artificial Intelligence Laboratory.
Her work addresses an increasingly important question for the technology industry: how can AI continue to advance without requiring ever-greater computing resources and energy consumption?
Improving the efficiency of existing infrastructure could help reduce the pressure to build additional data centers. However, the impact depends on how successfully research methods translate into real-world operations, and greater efficiency alone does not eliminate the environmental costs of expanding AI.
For Delimitrou, the challenge is not simply to develop more powerful technology. It is also to make computing systems more efficient, reliable, and sustainable as demand continues to grow.
See all the latest news from Greece and the world at Greekreporter.com. Contact our newsroom to report an update or send your story, photos and videos. Follow GR on Google News and subscribe here to our daily email!
