Researchers from the University of Ottawa, Columbia University, and Hanyang University are exploring how brain-inspired semiconductor chips could make artificial intelligence more energy-efficient. Their research reviews new materials and chip designs that mimic how the human brain processes information. The goal is to build faster, lower-power computing systems that could help reduce the growing electricity demands of AI data centers.
The research team reviewed advances in electronic materials, device physics, and circuit design to identify promising ways to build brain-inspired computing hardware. The findings were published in the journal Advanced Functional Materials in 2026.
Brain-inspired chips
Rather than reporting one new chip or a single laboratory experiment, the paper provides a roadmap for researchers working on neuromorphic computing. This approach aims to bring brain-like information processing into physical electronic devices.
The University of Ottawa, working with Columbia University and Hanyang University, contributed to the review of emerging semiconductor technologies. Lead author Dr. Kannan Udaya Mohanan, a postdoctoral fellow in Professor Chang-Hyun Kim’s research group at the University of Ottawa, helped examine the materials and designs that could support future brain-inspired chips.
Professor Chang-Hyun Kim, an associate professor in the university’s School of Electrical Engineering and Computer Science, leads its Semiconductor and Microelectronics Laboratory. The team assessed different approaches to building artificial neurons, which are electronic components designed to perform functions similar to neurons in the brain.
Brain-inspired chips aim to reduce processing tasks
Modern AI systems require large amounts of computing power to train and run complex models. In data centers, moving information between memory and processors can consume time and energy, creating a problem known as the “memory wall.”
Traditional chips usually separate memory storage from processing tasks. Brain-inspired designs aim to reduce some of this movement by changing how computing elements process and exchange information. If successful, these systems could help lower power use as AI workloads grow.
Neuromorphic computing is a technology that takes ideas from the way the brain processes information. Instead of relying entirely on conventional processor designs, it uses electronic circuits that imitate some functions of biological neurons and their connections.
The review examines emerging materials and device architectures that could serve as artificial neurons. These components may process signals through electrical responses that combine inputs and produce an output, much like simplified models of brain cells. Researchers are exploring ways to connect these devices into larger systems that can perform useful computing tasks.
If these designs can be developed at scale, they could help reduce the energy required for certain AI workloads. More efficient hardware could also lower operating costs for data centers and reduce pressure on electricity supplies.
The technology may have uses beyond large computing facilities. The researchers point to possible applications in lightweight wearable devices, personalized health technology, and systems that allow people to interact more naturally with machines. Some emerging materials may also support thin, flexible electronic devices.
The publication is a research review, not a demonstration that a new brain-inspired chip is ready for commercial use. The materials and designs discussed still face challenges involving manufacturing, reliability, integration, and scaling to large systems.
Neuromorphic chips may also be better suited to some tasks than others. Their real-world energy savings will depend on the device design, the workload, and how well the hardware works with existing computing systems.
As AI use expands, reducing the energy needed to process information is becoming an important technology goal. Brain-inspired chips offer a different way to approach computing by rethinking how electronic devices handle information, rather than relying only on improvements to traditional processors.
The review from the University of Ottawa and its partner universities gives researchers a guide to promising materials and architectures for future development. If these technologies can move from research into practical products, they could support more energy-efficient AI systems and smaller, smarter personal devices.
