We present a new perspective for the role of Termitomyces fungi in the mutualism with fungus-growing termites. According to the predominant view, this mutualism is as an example of agriculture with termites as farmers of a domesticated fungus crop, which is used for degradation of plant-material and production of fungal biomass. However, a detailed study of the literature indicates that the termites might as well be envisioned as domesticates of the fungus. According to the “ruminant hypothesis” proposed here, termite workers, by consuming asexual fruiting bodies not only harvest asexual spores, but also lignocellulolytic enzymes, which they mix with foraged plant material and enzymes of termite and possibly bacterial origin. This mixture is the building material of the fungus garden and facilitates efficient degradation of plant material. The fungus garden thus functions as an external rumen for termites and primarily the fungi themselves benefit from their own, and gut-derived, lignocellulolytic enzymes, using the termites to efficiently mix these with their growth substrate. Only secondarily the termites benefit, when they consume the degraded, nitrogen-enriched plant-fungus mixture a second time. We propose that the details of substrate use, and the degree of complementarity and redundancy among enzymes in food processing, determine selection of horizontally transmitted fungal symbionts at the start of a colony: by testing spores on a specific, mechanically and enzymatically pre-treated growth substrate, the termite host has the opportunity to select specific fungal symbionts. Potentially, the gut-microbiota thus influence host-fungus specificity, and the selection of specific fungal strains at the start of a new colony. We argue that we need to expand the current bipartite insect-biased view of the mutualism of fungus-growing termites and include the possible role of bacteria and the benefit for the fungi to fully understand the division of labor among partners in substrate degradation.