The malignant brain tumour glioblastoma multiforme (GBM) displays significant levels of cellular plasticity. It is often assumed that the tumour microenvironment (TME) regulates GBM cell states, but we know little about the spatial organisation of GBM tumours and TME-derived signals driving malignant cell state transitions. Here, I will present an integrated single cell and spatial multi-omic approach to resolve the tissue architecture of GBM. First, I will present cell2location, a new computational tool that can map fine-grained cell types in spatial transcriptomics data. Second, I will present GBM-space, a new collaborative effort to discover TME-GBM interactions using multi-modal genomics. Using joint single cell transcriptomic and epigenomic profiling, we define recurrent malignant cell states and expand the description of the TME cells. Using Visium spatial RNA-sequencing, we profile distinct tumour sites and observe significant regional heterogeneity of the GBM microenvironment. Finally, we integrate single cell and spatial transcriptomics using cell2location and discover that malignant cell states regionally segregate in GBM and associate with distinct TMEs. Our efforts reveal the novel spatial organisation of GBM tumours and identify putative TME signals that regulate malignant cell states.