Understanding the microbial and chemical diversities, as well as what affects these diversities, is important for modern manufacturing of traditional fermented foods. In this work, Chinese dark teas (CDTs) that are traditional microbial fermented beverages with relatively high sample diversity were collected. Microbial DNA amplicon sequencing and mass spectrometry-based untargeted metabolomics show that the CDT microbial $β$ diversity, as well as the nonvolatile chemical $α$ and $β$ diversities, is determined by the primary impact factors of geography and manufacturing procedures, in particular, latitude and pile fermentation after blending. A large number of metabolites sharing between CDTs and fungi were discovered by Feature-based Molecular Networking (FBMN) on the Global Natural Products Social Molecular Networking (GNPS) web platform. These molecules, such as prenylated cyclic dipeptides and B-vitamins, are functionally important for nutrition, biofunctions, and flavor. Molecular networking has revealed patterns in metabolite profiles on a chemical family level in addition to individual structures.