Artificial intelligence is rapidly transforming the world—but it’s also consuming enormous amounts of energy.
Data centers powering AI models rely on thousands of computer chips that generate intense heat.
To keep them from overheating and slowing down, massive cooling systems run constantly, using huge amounts of electricity. As AI demand grows, so does the urgency to develop technology that can process information faster while using far less energy.
Engineers at the University of Houston have created a breakthrough solution: a newly engineered ultra-thin film that could dramatically boost the performance of AI chips while slashing their power consumption.
Their findings, published in ACS Nano, introduce a cutting-edge two-dimensional dielectric film—a type of electrical insulator—that replaces traditional materials inside integrated circuits. Unlike conventional insulators, the new material does not store electric charge.
That means less energy wasted as heat and faster signal transmission inside the chip.
According to Professor Alamgir Karim, who led the research, AI’s rapid growth has pushed global energy needs to unprecedented levels.
Modern data centers must run powerful cooling equipment alongside high-performance chips, all to ensure fast processing speeds and long chip lifespans. Karim and his team set out to solve this problem at the material level.
Their innovation centers on creating what is known as a “low-k” dielectric. The “k” refers to the material’s permittivity—how much electrical energy it can store.
Traditional high-k materials store more charge, which leads to more heat. Low-k materials, by contrast, store less energy and therefore help chips run faster and cooler.
To design this new material, Karim and his former doctoral student, Maninderjeet Singh (now a postdoctoral researcher at Columbia University), turned to lightweight covalent organic frameworks—highly structured materials made mostly of carbon and other light elements.
These materials have porous, crystalline structures that allow electrical signals to move quickly with minimal interference. This helps reduce signal delays, power loss, and overheating.
The team fabricated the thin films using a method called synthetic interfacial polymerization. In this process, molecular building blocks are dissolved in two liquids that do not mix.
As the components meet at the boundary between the liquids, they stitch themselves together, forming strong, perfectly layered crystalline sheets. The method was originally pioneered by scientists including the 2025 Nobel Prize winner in Chemistry, Omar M. Yaghi.
The resulting 2D material has an ultralow dielectric constant and extremely high electrical breakdown strength, meaning it can withstand the high voltages required in advanced devices. It also remains stable at high operating temperatures—an essential quality for data centers and AI hardware.
Karim believes the new thin film could play a major role in reducing the energy demands of AI worldwide.
With the electronics industry urgently searching for ways to make chips more efficient, this innovation offers a promising path forward—not only for AI, but for all advanced computing technologies that power modern life.
Source: University of Houston.
