The physical foundations of tomorrow’s AI: what new electronics can change, and what the evidence actually supports.
From semiconductor scaling and emerging materials to memristors and photonic computing, this collection examines the devices behind computation. Each review follows one research direction and asks how it could change AI’s computing capacity, response time, energy demand and cost.
From a laboratory result to a useful system. A smaller device, a faster operation and a more efficient application are different achievements. Reviews distinguish experiments, simulations and projections, and explain what remains between a promising result and practical deployment.
Research directions
Ordered by publication date, newest first. Each review states when the evidence was examined; this is a collection of dated scientific assessments, not a live technology ranking.
-
Computing with light: what photonics could change for AI
Read the review: Computing with light: what photonics could change for AILight can carry information, but it can also perform selected calculations inside a purpose-built circuit. Photonic computing is moving from isolated optical operations towards experimental systems that run neural networks. This review examines what recent demonstrations actually establish, from programmable tensor processors to tiny task-specific optical structures and computing-centre trials. Its central question is what these advances could change for artificial intelligence: useful computing capacity, response time, energy demand and cost. The answer depends on the whole system, including memory, conversion, precision, calibration and data movement. Published classification results are distinguished from general AI capability; projected power budgets are separated from measured consumption. The article explains why a fast optical operation does not guarantee a faster application, and why silicon can remain part of a photonic future. It identifies credible opportunities for hybrid and specialised accelerators while setting out the evidence still needed before broad efficiency or economic claims can…
Reading the evidence
Selection is editorial and not exhaustive. Reviews connect physical principles, experimental results and system requirements to their possible consequences for AI. Primary scientific publications are preferred; a manufacturer claim, a peer-reviewed experiment, an independent reproduction and a projection are identified as different forms of evidence.
Results from different tasks or measurement boundaries are not combined into a synthetic ranking. Costs and performance gains are quantified only where the evidence permits, and an illustrative calculation is labelled as such. Newly published articles appear in this collection automatically.