Molecular-scale characterization and multi-omics techniques are producing multimodal datasets that reveal the structure and reactivity of soils, sediments, and organic matter at unprecedented resolution. Methods like high-resolution mass spectrometry, X-ray absorption spectroscopy, X-ray computed tomography, sequence-based approaches, metabolomics, and metaproteomics are enabling links between chemistry, pore structure, and biotic controls in biogeochemical models. Coordinated campaigns and platforms such as the Molecular Observation Network (MONet) are improving access to these datasets, but a major challenge remains: translating molecular-scale data into process-based and data-driven models across scales. This session seeks contributions addressing this gap through experimental, modeling, and computational perspectives, including scaling molecular observations, multimodal data fusion, AI/ML applications, and model–data integration. Researchers using open-source molecular datasets or pursuing novel analyses and synthesis are especially encouraged, with the goal of advancing predictive understanding of soil and subsurface biogeochemistry.