Abstract: Under the global goal of carbon neutrality, soil carbon sequestration—as a critical component of the terrestrial ecosystem carbon cycle—has garnered significant attention. Coal mining subsidence areas, as typical human-impacted ecosystems, are characterized by soil structural degradation, reduced fertility, and imbalances in microbial communities. Although the effects of artificial restoration on soil carbon sequestration potential and microbial community structure in these areas have been extensively studied, the evolutionary characteristics of soil microbial carbon sequestration potential during natural recovery remain poorly understood. To address this knowledge gap, this study selected coal mining subsidence areas as the research object and systematically collected soil samples from farmlands at various stages of natural recovery. High-throughput sequencing technology, structural equation modeling, and molecular ecological network analysis were comprehensively applied to investigate the dynamic changes in carbon-sequestering microbial community structures and their intrinsic relationships with soil carbon sequestration potential. Results indicated that natural recovery effectively restored the physicochemical properties and enzyme activities of farmland soils in coal mining subsidence areas. During the recovery process, the soil carbon sequestration potential was primarily influenced by Proteobacteria and Acidobacteria. As natural recovery progressed and soil damage gradually diminished, variations in pH and soil organic matter (SOM) content regulated the activities of catalase (CAT) and β-glucosidase (BG), thereby affecting the abundance of carbon sequestration-related genes in Proteobacteria and Acidobacteria, which in turn influenced the overall carbon sequestration potential. At the early stage of natural recovery, the interaction network between microbial communities involved in carbon sequestration and environmental factors was relatively simple, indicating low ecosystem stability. However, as recovery advanced, the complexity and stability of the system increased, evidenced by a growing number of network nodes, stronger inter-node connections, and higher vertical connectivity and clustering coefficients. This study, for the first time, systematically elucidated the evolutionary patterns and regulatory mechanisms of soil carbon sequestration potential in farmland soils during the natural recovery of coal mining subsidence areas. These findings provide new insights into the comprehensive evaluation of ecological benefits associated with natural recovery processes. The results not only enrich the theoretical foundation of ecological restoration in coal mining subsidence areas but also offer crucial scientific support for achieving China's carbon neutrality goals. Furthermore, they hold significant practical value for guiding sustainable land use and ecological restoration efforts in such regions.