<p>Brain metastasis remains a major clinical challenge because intracranial progression and therapeutic resistance arise not only from tumour-intrinsic programmes but also from coordinated communication across molecular, cellular, spatial and systemic scales. In this Review, we integrate findings from spatial transcriptomics/proteomics, single-cell and multimodal omics, and computational systems modelling to frame brain metastasis as a multiscale communication disorder in which immune, glial, vascular and systemic networks collectively shape tumour-permissive ecosystems. We discuss how network-based analyses identify communication hubs that can be translated into therapeutic hypotheses and provide a rationale to guide drug repurposing and rational combination strategies targeting tumour–microenvironment interactions. We further examine how multiscale modelling and artificial-intelligence-enabled integration of spatial and molecular data may enable patient stratification, response prediction, and adaptive trial designs, while highlighting key barriers, including data harmonization, spatial resolution limits, model interpretability and clinical deployment constraints. Reframing brain metastasis as a multiscale communication disorder provides a unifying conceptual and methodological foundation for therapies that disrupt metastatic ecosystems and improve durable intracranial disease control.</p>