Applications of AI in VLSI Design: A Comprehensive Revie

Abstract

This paper presents a comprehensive review of the rapidly expanding role of AI in transform
ing the VLSI design ecosystem. As technology nodes advance toward 3 nm and beyond, traditional EDA
tools struggle with escalating complexity, rising verification costs, and increasingly interdependent power, per
formance, and area (PPA) constraints. The review examines how AI methodologies-particularly Deep Rein
forcement Learning (DRL), supervised learning, and Graph Neural Networks (GNNs)-address these challenges
across the complete VLSI design flow. AI-driven approaches are shown to significantly enhance RTL optimiza
tion, autonomous Design Space Exploration (DSE), physical design stages such as floor planning, placement,
routing, and timing closure, as well as verification and testing processes. The paper further highlights AI’s
expanding impact in analog and mixed-signal design through variation-aware modeling and intelligent simu
lation reduction. Industrial adoption is validated through case studies demonstrating substantial productivity
gains and measurable PPA improvements delivered by commercial AI-enabled EDA platforms. Emerging
research directions-including Circuit Foundation Models (CFMs) and Explainable AI (XAI)-are identified as
critical enablers for scalable, interpretable, and fully autonomous design flows. Overall, the review underscores
AI’s pivotal role in shaping the next generation of high-performance, energy-efficient semiconductor design
methodologies.