Abstract: As China enters a phase of stock-based planning and quality-oriented urban renewal, accurate identification of renewal potential is central to megacity governance. Drawing on spatial gene theory, this study treats building morphology as a phenotypic expression and develops an artificial-intelligence framework for gene identification and renewal prediction. Shanghai building-rooftop data are used to derive morphological indicators, and a Gaussian mixture model identifies seven spatial-gene types. Parcels changing from the old-urban type to the central-city type are defined as renewed areas. A random forest model predicts renewal with fourteen morphological, locational, and socioeconomic variables. Renewal is concentrated near the suburban and outer rings and shifts from north toward southwest. Central districts, western Shanghai, and old suburban centers show high potential. Total building area, mean perimeter, construction age, centrality, and housing price are key factors.