The widespread application of artificial intelligence is reshaping the discipline of geography, yet the deeper challenge lies in how the computational thinking embodied in AI can be deeply integrated with geospatial thinking. Drawing on the outcomes of the“spatial computational thinking”model and the“Encoding Geography”initiative, this paper constructs a theoretical framework of“dual-thinking integration”of computational thinking and geospatial thinking for geography education in universities. The framework reveals the diff erences and complementarities between the two types of thinking across four dimensions: epistemology, methodology, capability, and innovation. Furthermore, it proposes a three-stage teaching pathway—“geographic problem formulation, computational method implementation, and geospatial meaning reconstruction”—aimed at guiding students towards the synergistic operation of the two kinds of thinking through iterative dialogue between output results and spatial interpretation. A case study based on an economic geography module titled “Industrial Spatial Agglomeration”illustrates that this pathway helps students move from merely“operating tools” to“thinking geographically with tools,”thereby offering a feasible approach to addressing the challenges of cultivating geography talents in the context of rapid technological iteration.