Climate change is altering the frequency and intensity of precipitation events. However, the impact of this changing hydroclimate on the functional diversity of soil microbial communities still needs to be clarified. Addressing this question is important for quantifying the fate of soil organic matter under climate change, whereas the key knowledge gap is to understand how past and current hydroclimate events regulate various functions of microbial communities over time and the resulting impact on SOM decomposition. To bridge this knowledge gap, we quantified the properties of past and current drought events and integrated them with soil thermal and physiochemical properties as well as omics-based microbial functional information from multiscale observation networks to predict temporal dynamics of soil microbial functions involved in soil carbon (C) and nutrient cycles. We have utilized machine learning models to interpret the regulations of long-term average climate and physiochemical conditions on soil microbial functional diversity and leveraged these relationships to predict the spatial distribution of microbial functional composition across the continental United States (CONUS) scale. Furthermore, we employed sequence-based deep learning models to predict the temporal dynamics of soil microbial functional diversity and its response to environmental changes. Building on these studies, we further identified the temporal distribution of soil microbial function in response to current and previous drought events. Our studies indicated that the temporal dynamics of these microbial functions involved in C and nutrient cycling were strongly dominated by the current and historical drought events. Our findings provided gene-scale evidence and the predictable model to elucidate the life history strategies of soil microbial communities under changing hydroclimate.