We are pleased to announce the publication of our perspective paper, “Perspective: Magnetic Tunnel Junctions for In-sensor Computing”, in Nano Futures.
The paper reviews recent advances in MTJ-based reconfigurable Boolean logic, logic-in-memory, and neural-network primitives for feedforward, convolutional, recurrent, and spiking neural networks. It also discusses emerging magnetic near-sensor computing, multifunctional MTJs, and voltage-controlled MTJ processing-in-pixel concepts, while identifying key challenges in developing scalable sensor-computing arrays. The perspective highlights a major future opportunity: engineering large-area MTJ sensor arrays that can perform local thresholding, event detection, feature extraction, and compressed readout directly at the magnetic sensing interface.