Water scarcity is a major challenge facing many regions worldwide, especially arid and semi‐arid areas that are increasingly vulnerable to climate change. This study aimed to project water availability in the Amu Darya Basin (ADB) of Central Asia under four Shared Socioeconomic Pathways (SSPs) from the Coupled Model Intercomparison Project Phase Six (CMIP6) during two upcoming periods (2020–2059 and 2060–2099). The study used a robust machine learning approach, namely a Random Forest (RF) model, to simulate Gravity Recovery and Climate Experiment (GRACE) Terrestrial Water Storage (TWS) data from precipitation and maximum and minimum temperatures (Tmax and Tmin). It then incorporated precipitation, Tmax and Tmin from four selected CMIP6 GCMs, into a water storage model to project spatiotemporal changes in water availability across the basin. The study also evaluated the relative impacts of land use and population on TWS. Results indicate an increase in TWS by approximately 4 cm in the basin's eastern, northwestern and southwestern regions in both future periods, while a decrease by approximately −4 cm in the remaining areas. These projections suggest that TWS will decline in densely populated regions and increase in certain intensively cultivated areas. The most pronounced increase in TWS is anticipated in the snow‐covered Tundra climate zone of the basin. This is attributed to the melting of glaciers, which contributes to runoff in the tributaries of the Amu River. The findings highlight the importance of considering climate change and socioeconomic factors when projecting water availability in arid and semi‐arid regions. The projected changes in TWS have important implications for water resources management in the ADB, particularly in densely populated and intensively cultivated areas.