Scalable neuromorphic 2D molybdenum disulfide memristors enable in-memory computing and operate at significantly lower voltages

https://phys.org/news/2026-02-2d-memristors-ai-energy-problem.html

"performs vector-matrix multiplication, mathematical backbone of AI, directly within hardware at fraction of energy cost... can be stacked or integrated into existing CMOS (silicon) technology, offering a viable path to scaling AI hardware without requiring massive new power plants... suitable for edge AI, allowing complex AI models to run locally on small devices (like smartphones or sensors) for extended periods without needing a constant connection to a power-hungry data center"

Related:

SKKU develops hafnium oxide-based next-generation memory devices for AI hardware
https://www.eurekalert.org/news-releases/1119575

Highly stable self-rectifying memristor arrays: Enabling reliable neuromorphic computing via multi-state regulation
https://www.eurekalert.org/news-releases/1118280