With conventional computer chips approaching physical limits and AI workloads rising, researchers are turning to the brain for design ideas. Energy use from AI data centers is projected to double by the end of the decade, prompting work on more efficient hardware.
Neuromorphic computing aims to process information more like biological neural networks. "We’re trying to make devices that behave more like the brain itself," says Suchi Guha, a professor of physics at the University of Missouri and a core faculty member at the Materials Science and Engineering Institute. Guha highlights the brain’s efficiency: it performs complex tasks using about 20 watts of power, roughly the same as an old light bulb.
Guha’s team builds synapse-like electronic components using organic transistors that combine memory and processing in one place. Conventional chips move data between separate memory and processors, which both slows operations and consumes energy. In contrast, synaptic devices can perform and store computation locally, reducing that energy cost.
The researchers tested several organic materials that looked almost identical on the surface but behaved very differently as synaptic transistors. They concluded that the interface—the thin boundary where the semiconductor meets an insulating layer—was the key factor. By clarifying how molecular design and interface quality affect synaptic behavior, the work offers guiding principles for neuromorphic hardware that could enable brain-like AI to learn more efficiently and use far less power.
- Pattern recognition
- Decision-making
- Learning with low energy
The study appears in ACS Applied Electronic Materials. Additional coauthors are from Mizzou and Hamad Bin Khalifa University. Source: University of Missouri.
Difficult words
- neuromorphic — design inspired by biological brains for computing
- synaptic — relating to synapses between nerve cells
- transistor — small electronic switch that controls currenttransistors
- interface — thin boundary where two materials meet
- semiconductor — material that partly conducts electricity
- insulating — preventing or reducing electrical current flow
- molecular — relating to molecules, very small particles
- workload — amount of computing tasks or data loadworkloads
Tip: hover, focus or tap highlighted words in the article to see quick definitions while you read or listen.
Discussion questions
- How might neuromorphic hardware change the energy use of AI in data centers? Give reasons from the article.
- What challenges do you think engineers face when trying to make devices behave like the brain? Use points from the text to support your ideas.
Related articles
AI audio summaries of research can help — and err
Researchers tested Google’s NotebookLM, which turns research papers into podcast-style audio. The summaries were engaging and clearer for teaching, but every audio overview contained mistakes, so the authors advise reading the original papers to check claims.
Band of Holes at Monte Sierpe: an Indigenous accounting system?
New evidence suggests the Band of Holes at Monte Sierpe in southern Peru was part of a pre‑European Indigenous system for accounting and exchange. Researchers used sediment analysis and drone images to reach this conclusion.
