Grassland degradation has become a serious environmental problem worldwide. However, the link between grassland degradation and soil functional microorganisms remains unclear. Here, we investigated the effects of subalpine meadow degradation on the taxonomic and functional communities of the soil microbiome on Mount Wutai on the eastern margin of the Loess Plateau, China, using shotgun metagenomic sequencing. The results showed that vegetation degradation decreased most soil nutrients (e.g., soil organic carbon [SOC, −48.9%], total nitrogen [TN, −15.2%] and carbon [TC, −38.3%]), leading to significant changes in the taxonomic (r = 0.892, p < 0.001) and functional compositions (r = 0.972, p < 0.001) of the microbial communities (explaining 66.6% and 64.5% of the variance, respectively). The structural equation model (SEM) showed that variations in the soil nutrient status and plant parameters caused by degradation resulted in changes in the microbial community structure and function. The variations in the plant and microbial communities during degradation altered carbon (C) and nitrogen (N) cycling separately or interactively. Among the C degradation‐related genes, the relative abundances of cellulose degradation, sugar utilization and chitin degradation genes in moderately degraded (MD) and heavily degraded (HD) meadows were significantly higher than those in nondegraded (ND) meadows (p < 0.05). The abundance of denitrification genes (nirK, nirS, norB and norC) and nitrification genes (pmoA/amoA, pmoB/amoB, pmoC/amoC and hao) was higher in all degraded meadows, which increased the risk of soil N loss. Our findings revealed the negative feedback of microbes due to subalpine meadow degradation, during which metabolically active microbes tended to accelerate soil nutrient loss.