Reducing the AI Energy Crisis: UC Merced Explores Topological Materials for Sustainable Computing
Artificial intelligence is transforming every facet of modern life, but its growth comes with a staggering environmental cost. As LLMs become more integrated into our daily workflows, the electricity required to power them is skyrocketing. To combat this, researchers at UC Merced are diving into the world of topological materials to build a more energy-efficient foundation for next-generation computing.
- A single ChatGPT query consumes roughly 0.34 watt-hours, which is about 10 times the energy of a standard Google search.
- U.S. Data centers consumed 183 terawatt-hours of electricity in 2024, exceeding 4% of the country’s total consumption.
- UC Merced is utilizing a $6 million systemwide grant to research topological materials that allow electrons to move in unusual ways, potentially reducing hardware energy needs.
- The project is a multidisciplinary effort involving five University of California campuses and two national laboratories.
The Massive Energy Footprint of Modern AI
It’s no secret that AI is power-hungry. The scale of energy consumption is becoming a critical hurdle for sustainable tech growth. According to data from the Pew Research Center, U.S. Data centers used 183 terawatt-hours of electricity in 2024. To put that in perspective, that’s more than 4% of the entire United States’ electricity consumption—an amount roughly equivalent to the annual electricity demand of Pakistan ([UC Merced]).
The inefficiency is most apparent when comparing traditional search to generative AI. A standard ChatGPT query consumes approximately 0.34 watt-hours, making it roughly 10 times more energy-intensive than a typical Google search.
Topological Materials: The Path to Efficient Hardware
To solve this energy crisis, we can’t just rely on better software; we need a hardware revolution. This is where topological materials come in. Led by Chemical and Materials Engineering Professor Elizabeth Nowadnick, a team at UC Merced is investigating these unique materials where electrons move in unusual ways ([PublicNow]).
These materials could provide the blueprint for more efficient computing platforms. By changing how electrons behave at the hardware level, researchers hope to create systems that process massive amounts of data without the current energy overhead.
A Multidisciplinary, Systemwide Effort
This isn’t a solo mission. The UC Office of the President issued a $6 million grant to fund a systemwide effort to develop more efficient computing. Of this total, UC Merced received $810,000 to lead its specific research arm. The project is a massive collaborative effort involving:
- University of California Campuses: Berkeley, Irvine, Merced, San Diego, and Santa Barbara.
- National Laboratories: Lawrence Livermore and Los Alamos.
Putting AI to Work on Its Own Problems
In a poetic twist, the researchers aren’t just fighting AI’s energy consumption—they’re using AI to solve it. Professor Nowadnick’s team has two primary goals: use AI to accelerate the discovery of topological materials and create a framework that can be applied to other functional materials, such as superconductors, magnets, and related quantum materials ([UC Merced]).
By leveraging AI to find these materials faster, the university aims to keep the U.S. At the forefront of economic and scientific leadership in emerging tech.
Beyond Hardware: UC Merced’s Broader AI Integration
While the hardware research targets the energy crisis, UC Merced is simultaneously integrating AI into its academic and administrative infrastructure. The university established the UC Merced AI Advisory Council on February 11, 2026, to guide the responsible integration of these tools ([UC Merced AI]).
The campus is also exploring practical applications, such as using Amazon Bedrock LLMs to automate research data extraction from thousands of declassified ARPA documents and negotiating ChatGPT EDU licenses for the campus community.
Frequently Asked Questions
Why are topological materials better for AI?
These materials allow electrons to move in unusual ways, which can lead to more efficient platforms for computing, potentially reducing the heat and electricity required to run complex AI models.
How much more energy does AI use compared to search?
A standard ChatGPT query uses about 0.34 watt-hours, which is approximately 10 times more than a Google search.
Who is funding the UC Merced research?
The project is funded by a $6 million grant from the UC Office of the President, with $810,000 specifically allocated to UC Merced.
Looking Ahead
The trajectory of AI is clear: it will continue to grow in capability and adoption. But, for that growth to be sustainable, the industry must move beyond simply adding more GPUs and data centers. The work being done at UC Merced and its partner institutions represents a critical shift toward “green” hardware, ensuring that the digital breakthroughs of tomorrow don’t come at the expense of the planet’s energy stability.