IEEE Spectrum AI

New Devices Might Scale the Memory Wall

Researchers at UC San Diego unveiled a new bulk resistive RAM (RRAM) that switches an entire material layer instead of forming filaments, enabling 3D stacking and selector‑free operation. The devices are 40 nm wide, can be stacked in up to eight layers, and each cell offers 64 resistance states in the megaohm range. A 1‑kilobyte array of these stacks ran a continual‑learning algorithm on wearable sensor data, achieving 90 % classification accuracy comparable to digital neural networks. The breakthrough promises on‑chip AI inference and learning for edge devices, though high‑temperature data retention remains uncertain.