Memristor-based analog-to-digital converter overcomes bottlenecks, significantly enhancing how AI chips process signals

https://techxplore.com/news/2025-12-memristor-based-boosts-energy-efficiency.html

"AI accelerators bulky/ power-hungry, overcome using memristors dynamically fine-tuning signal conversion based on incoming data rather than fixed, inefficient settings... 15.1X improved energy efficiency, 12.9X reduced circuit area allowing for smaller chips... Compute-In-Memory systems: 57% reduced total chip energy, 30% reduced overall size... high accuracy across various neural network models, faster, operates on much smaller batteries... edge devices: smartphones, sensors"

Related: Atom-thin, content-addressable memory enables edge AI applications
https://phys.org/news/2026-01-atom-thin-content-memory-enables.html