Climate change globally affects soil microbial community assembly across ecosystems. However, little is known about the impact of warming on the structure of soil microbial communities or underlying mechanisms that shape microbial community composition in subtropical forest ecosystems. To address this gap, we utilized natural variation in temperature via an altitudinal gradient to simulate ecosystem warming. After 6 years, microbial co‐occurrence network complexity increased with warming, and changes in their taxonomic composition were asynchronous, likely due to contrasting community assembly processes. We found that while stochastic processes were drivers of bacterial community composition, warming led to a shift from stochastic to deterministic drivers in dry season. Structural equation modelling highlighted that soil temperature and water content positively influenced soil microbial communities during dry season and negatively during wet season. These results facilitate our understanding of the response of soil microbial communities to climate warming and may improve predictions of ecosystem function of soil microbes in subtropical forests. Climate change globally affects soil microbial community assembly across ecosystems. We utilized natural variation in temperature via an altitudinal gradient to simulate ecosystem warming. After 6 years of experiment, microbial co‐occurrence network complexity increased with warming, and changes in their taxonomic composition were asynchronous due to contrasting community assembly processes. Structural equation modelling highlighted that soil temperature and water content positively influenced soil microbial communities during dry season and negatively during wet season.