Many studies suggest a positive and monotonic relationship between technological progress and wage income inequality since 1980s for industrialized economies. We examine this topic in the context of the Chinese economy where new technologies like automation, AI and digitalization have witnessed worldís most rapid growth in the last decade. Surprsingly, we Önd an inverted U-Shaped relationship - a "Digital Kuznets Curve", using a panel dataset constructed in line with the newly published "2021 Categorization of Core Industries of Digital Economy". We then set out a task-based growth model with heterogenous human capital and occupational choice, and show that this hump-shaped relationship can emerge either by introducing an erosion e§ect of digitalization on worker ability that increasingly counterveils the skill-biasing e§ect, or by directly adding a dynamic learning cost that captures the externality of digitalization. Our study contributes to the understanding of the nature of digitalization in re-shaping labor market structure.