The continued miniaturization of field‐effect transistors (FETs) has reached a stage where geometric scaling alone can no longer deliver the necessary performance, power, and area improvements of nanoelectronic circuits. This article presents a forward‐looking, AI‐guided framework for the co‐optimization of advanced FETs, integrating data‐driven materials discovery, physics‐informed device modeling, and multiscale fabrication design. We conceptually decompose FET development into three tightly coupled domains—material design, device design, and fabrication design—and discuss how emerging machine learning, generative modeling, and physics‐informed surrogate models could enable more agile, data‐driven exploration and design within each. We further discuss the potential of hierarchical reinforcement learning to orchestrate cross‐level decision‐making and efficiently navigate the high‐dimensional, nonconvex design landscape inherent to advanced transistor technologies. The proposed framework illustrates a possible pathway toward intelligent, autonomous, and sustainability‐conscious semiconductor design. By articulating this vision, we aim to stimulate discussion and inspire future research toward bridging materials, devices, and manufacturing in the subnanometer regime.
Nath et al. (Fri,) studied this question.