Uncovering the tissue molecular architecture at single-cell resolution could help better understand the biological and pathological processes in organisms. However, bulk RNA-seq can only measure gene expression in cell mixtures, without revealing molecular heterogeneity and spatial patterns of single cells. Herein, we introduce Bulk2Space (https://github.com/ZJUFanLab/bulk2space), a deep learning framework-based spatial deconvolution algorithm that can simultaneously reveal the spatial and cellular heterogeneity of bulk RNA-seq data. The use of bulk transcriptomics to validate Bulk2Space revealed, in particular, the spatial variance of immune cells in different tumor regions, the molecular and spatial heterogeneity of tissues during inflammation-induced tumorigenesis, and spatial patterns of novel genes in different cell types. Moreover, Bulk2Space further annotated cell types that cannot be identified by original methods when it was used to spatially deconvolve the bulk data derived from our in-house developed sequencing approach Spatial-seq. Bulk2Space is distributed via an open-source Python software package.