New Brain Chip for Faster, Smarter, Low-Power Computing

A new microchip from Cornell processes data and wireless signals at lightning speed, using a neural network- while sipping tiny amounts of power.

Cornell University researchers created a low-power microchip known as a “microwave brain,” which is the first processor to compute on both ultrafast data streams and wireless communication signals using microwave physics.

What Is the ‘Microwave Brain’ Chip?

The processor, described in detail in the journal Nature Electronics, is the first genuine microwave neural network to be fully integrated on a silicon microchip. It does real-time frequency domain computation for tasks such as radio signal decoding, radar target tracking, and digital data processing while using less than 200 milliwatts of power.

How Does the Microwave Neural Network Work?

“Because it can distort in a programmable way across a wide band of frequencies instantly, it can be repurposed for several computing tasks,” said lead author Bal Govind, a Ph.D. student who conducted the research with Maxwell Anderson, another doctorate student. “It bypasses a large number of signal processing steps that digital computers normally have to do.”

That functionality is possible by the chip’s design as a neural network, a computer system based on the brain that employs interconnected modes formed by tunable waveguides. This enables it to see trends and learn from data.

Major Advantages of Microwave Neural Network

However, unlike standard neural networks, which rely on digital operations and step-by-step instructions timed by a clock, this network employs analog, nonlinear behavior in the microwave realm, allowing it to handle data streams in the tens of gigahertz – significantly faster than most digital circuits.

How Microwave Neural Network Outperforms Conventional Digital Circuits

“Bal threw away a lot of conventional circuit design to achieve this,” said Alyssa Apsel, professor of engineering and co-senior author alongside Peter McMahon, associate professor of applied and engineering physics. “Instead of trying to mimic the structure of digital neural networks exactly, he created something that looks more like a controlled mush of frequency behaviors that can ultimately give you high-performance computation.”

Features of Microwave Neural Network

The device can do both low-level logic processes and more complicated jobs, such as detecting bit sequences or counting binary values in high-speed data. It achieved 88% or higher accuracy on numerous classification tests involving wireless signal types, which is comparable to digital neural networks but requires a fraction of the power and size.

Paving the Way for Next-Gen Tech

The chip’s high sensitivity to inputs makes it ideal for hardware security applications such as recognizing irregularities in wireless communications across several microwave frequency bands, according to the researchers.

References:

1. An integrated microwave neural network for broadband computation and communication
   (Govind, B., et al. (2025). An integrated microwave neural network for broadband computation and communication. Nature Electronics. doi.org/10.1038/s41928-025-01422-1)

Source-Medindia