Catenaa, Wednesday, October 22, 2025- Cornell University engineers have developed a groundbreaking “microwave brain” chip that processes data using electromagnetic waves instead of conventional digital circuits, potentially transforming artificial intelligence and communications technology.
The chip, described in Nature Electronics, performs radar tracking, signal decoding, and data analysis at microwave frequencies while using less than 200 milliwatts of power.
Researchers say it represents the first fully functional microwave neural network integrated on a silicon chip.
Unlike traditional processors that rely on digital clock speeds, the device uses tunable waveguides that mimic brain-like neural connections. This design allows data to be processed in real time at tens of gigahertz — far beyond the reach of most digital systems.
Doctoral student Bal Govind, who led the project with researcher Maxwell Anderson, said the chip’s programmable distortion capabilities let it adapt to multiple computing tasks without the complex signal processing required by digital hardware.
The experimental chip achieved accuracy rates above 88% in wireless signal recognition tests while using a fraction of the power of digital neural networks.
Researchers say this combination of speed and energy efficiency could pave the way for on-device AI in wearables, smartphones, and satellites.
Supported by DARPA and the Cornell NanoScale Science and Technology Facility, the project’s next phase will focus on improving scalability and integrating the chip into hybrid microwave-digital computing systems.
