Projection of future woody cover is essential to understand potential changes in structure and functioning of terrestrial ecosystems. Previous studies mapped woody cover during historical periods observed by satellites, however, it remains unclear how woody cover is expected to change in response to future climate change. Here, we develop data-driven models to predict woody cover in Africa using multiple environmental predictors and show that woody cover can be accurately modeled using Random Forest. Empirically-based simulations forced by precipitation from CMIP6 project an overall increase in woody cover at the continental scale by 2100. However, increases are mainly occurring in regions with annual precipitation less than ∼1,600 mm y⁻¹, whereas woody cover is predicted to decrease in areas of higher rainfall. Our results suggest that climate change may alter the functioning of dryland ecosystems by continued woody encroachment and cause a loss of carbon stocks in humid areas.