Alpine meadow degradation on the northeastern Qinghai-Tibet Plateau threatens the stability of semi-arid alpine ecosystems, yet how microbial structure and function respond remains unclear, impeding restoration. Here we investigated four alpine meadow sites in Maqu County (Gansu, China) spanning a degradation gradient (ND-SD). We measured vegetation attributes, soil physicochemical properties, microbial biomass (SMBC, SMBN), soil respiration (SR), metabolic quotient (qCO2), extracellular enzyme activities, and bacterial (16S rRNA) and fungal (ITS) communities using high-throughput sequencing. The results show that: (1) With increasing degradation, soil water content (SWC), soil organic carbon (SOC), soil total nitrogen (STN), and soil total phosphorus (STP) declined, whereas pH and electrical conductivity increased; microbial biomass, SR, and most enzyme activities decreased. (2) Microbial alpha-diversity (Shannon) also declined, and community composition shifted despite relatively stable dominant phyla. Several genera (notably Pseudomonas, Gp6, Sebacina, and Cryptococcus) emerged as stage-sensitive indicators and were significantly associated with enzyme activities and qCO2. (3) Structural equation modelling further revealed contrasting controls: bacterial community variation was mainly mediated by vegetation-driven changes in soil conditions, whereas fungal composition was more closely linked to soil properties and microbial biomass/substrate continuity. Together, these results show that microbial communities provide sensitive indicators of semi-arid alpine meadow degradation. Restoration should combine grazing control with rebuilding soil moisture and organic matter, while monitoring key taxa and functional metrics to guide recovery.