Micro-sized Chips Process Data Like Brain

The processor uses electromagnetic waves instead of digital logic, combining signal analysis and data processing within a single low-power chip architecture.

The Cornell microwave brain on a chip

Engineers develop a new kind of micro-sized chip which uses microwaves to process data. The chip functions as a neural network in the microwave domain and completes real-time tasks such as radar tracking, wireless signal decoding, and high-speed data processing. It operates with less than 200 milliwatts of power, showing an alternative path for efficient computing.

Instead of processing data through binary logic and timed instructions, the chip manipulates electromagnetic wave patterns within reconfigurable waveguides. Each mode acts like a node connected in the brain. Together, they identify patterns, learn from variations, and adapt to new signals in real time. This allows the processor to compute across tens of gigahertz of bandwidth, a range that exceeds most digital systems.

The study, published in Nature Electronics, presents the first microwave-based neural processor fully integrated on a silicon chip. It performs computations directly in the frequency domain, reducing the need for multiple digital processing steps that are common in conventional systems. This method enables faster data handling at lower energy cost.

The team from Cornell University demonstrates that the chip performs tasks such as binary pattern recognition and wireless signal classification with accuracy levels above 88 per cent. These results match digital neural networks in performance while using a fraction of their space and energy. The processor’s probabilistic design also makes it effective for complex tasks without the additional hardware usually needed for precision or error correction.

Researchers are now exploring future security and sensing devices that monitor radio communication bands on computing platforms, such as wearable or mobile devices. The work represents a shift toward merging analogue microwave physics with modern computing architectures for scalable on-device intelligence.