Machine Learning Computer-Aided Design of Electronic Devices and Circuits

Driven by the miniaturization of single components, which has reached few nanometers, chips can nowadays accommodate hundreds of billions of transistors. As a result, modern nanoelectronics devices and systems are becoming extremely complex. To remain truly effective, computer-aided design (CAD) tools must evolve in parallel with these advancements, keeping pace with the increasing demands of scale and functionality. In recent years, machine learning (ML) has seen an incredible development in several areas. In nanoelectronics, ML is already offering promising technological solutions that improve performance and provide predictions across the entire electronic design stack. From material properties to quantum transport and circuit behavior, ML techniques are making simulation workflows much more efficient and flexible, leading to a new paradigm of CAD: Machine Learning Computer-Aided Design. This review examines the latest developments in computer simulations for assessing the performance of electronic systems and the types of models used across the different levels of abstraction. We also discuss the challenges that still need to be addressed, especially the need for physics-based ML models, and explore future directions toward generative machine learning frameworks tailored specifically for electronic system design.