Processing maps in metal additive manufacturing have traditionally been developed to guideprocess design and ensure dense components free of defects such as keyholing, lack of fusion, and balling. However, these maps offer limited insight into the resulting microstructural characteristics of additively manufactured metals, including grain size, crystallographic orientation, and other structural features. One particularly important descriptor is the degree of microstructural heterogeneity introduced during additive manufacturing. In this work, we aim to understand, quantify, and predict microstructural heterogeneity in metal additive manufacturing. We focus on laser powder bed fusion of 316L stainless steel as a representative process–material system. Within the optimal processing window, we perform detailed electron microscopy analyses to characterize microstructural heterogeneity and map its evolution as a function of laser power and scan velocity. Complementary Kinetic Monte Carlo (KMC) simulations are carried out to model the resulting microstructure and compare with experimental observations. Both experiments and simulations reveal that increasing laser power and reducing scan speed generally promote grain growth. However, the degree of heterogeneity exhibits a non-monotonic trend, with a distinct maximum that depends on the specific power–velocity combination. The underlying mechanisms governing this behavior are discussed.