U of M awarded $1.2M to help improve energy efficiency for computer systems using AI
The investment from NSF will drive the future of semiconductors
MINNEAPOLIS / ST. PAUL (11/21/2024) — The University of Minnesota will receive more than $1.2 million over the next three years from the National Science Foundation’s Future of Semiconductors (NSF FuSe2) to advance semiconductor research and innovation. The project will focus on increasing the energy efficiency for computer systems that use artificial intelligence (AI).
Energy use for computing continues to increase with the growth of AI and machine learning. Modern AI requires large data centers that house thousands of specialized computers that are used to train AI models and generate responses. These data centers require a large amount of electricity and generate so much heat that extensive cooling systems are needed to prevent the computers from overheating. With the growth of AI, it will dramatically increase the need for large data centers.
The most recent report from the International Energy Agency shows that an AI-enabled ChatGPT query uses almost 10 times more energy than a standard Google search. The report also predicts exponential growth in the AI industry, increasing AI’s electricity demand by at least 10 times from 2023 to 2026.
With this new grant, University of Minnesota researchers will investigate ways to fundamentally redesign the computing systems that support AI, from the semiconductor materials and devices to the computing circuits and architectures.
"One of the exciting aspects of this project is the interdisciplinary co-design approach,” said Sarah Swisher, University of Minnesota associate professor in the Department of Electrical and Computer Engineering and one of the lead researchers on the grant. “Rather than one research group working on a new memory device and another group working separately on the computing circuit that relies on that memory device, we have assembled a team that brings those very different skills to the table to work together. By learning from each other, we will be able to more rapidly converge on the breakthrough solutions in each area that are needed to enable drastic reductions in the energy for computing."
The goal is to decrease the amount of energy needed to run these types of systems, which could reduce the strain on the electricity grid and mitigate the impacts of AI on climate change.
The University of Minnesota’s Technological Leadership Institute and the Minnesota Nano Center recently announced the addition of new training modules to help build the next generation semiconductor workforce through the Minnesota Semiconductor Manufacturing Consortium.
“In addition to the scientific research, we will be creating an innovative plan to strengthen the semiconductor workforce in Minnesota,” Swisher said. “We have a strong microelectronics industry here, but with the growth from the U.S. CHIPS and Science Act investments, we are simply going to need to train more people and strengthen that pipeline.”
Swisher said the University will use mentorship programs and hands-on semiconductor fabrication experience to attract students and prepare them for semiconductor jobs.
The University of Minnesota is home to leading semiconductor research, particularly in sensors, materials science, nanotechnology, spintronics, and an area called 3D heterogeneous integration—which allows the integration of analog and digital circuit blocks with greater speed and lower power consumption.
Earlier this year, the University named Steven Koester, a professor in the Department of Electrical and Computer Engineering and director of the Minnesota Nano Center as the University’s first Chief Semiconductor Officer and launched a website devoted to semiconductor and microelectronics research and education, chips.umn.edu.
Read more about the new grants to support the future of semiconductors on NSF’s website.