Microbiomes drive biogeochemical cycling across Earth's critical zones (soils, oceans, rivers, sediments, deep subsurface), yet our ability to translate genomic data into the ecologically and biogeochemically relevant traits that determine microbiome function remains limited. Large fractions of environmental microbial communities have never been cultured, and conventional metagenomic annotation captures only a minority of the complex traits governing element cycling. Emerging AI-enabled frameworks,  including genomic foundation models, protein structural predictions, and knowledge graph architectures, offer a path toward accurate, genome-informed trait inference across the full diversity of environmental microbiomes, enabling mechanistically grounded models of biogeochemical processes across Earth system scales. We invite submissions addressing microbial trait inference, trait-based biogeochemical modeling, and the application of machine learning to link microbial genomics to elemental cycling across any Earth system environment.