Land use and land cover change (LUCC) can exacerbate cloud-mediated climate warming. However, the long-term response of cloud cover to LUCC remains underexplored, particularly regarding differences between idealized and realistic forest expansion scenarios, as well as between forest and bioenergy expansion. Here, using simulations from the fully coupled Community Earth System Model (CESM), we demonstrate that large-scale idealized afforestation and bioenergy expansion accelerate the loss of low- and mid-level clouds while enhancing high-level clouds, thereby amplifying regional warming hotspots through intensified positive shortwave cloud radiative forcing. In contrast, realistic afforestation yields a net cooling effect. Idealized afforestation drives a pronounced decline in low-level cloud cover (1.14 times globally, 1.52 times over land), followed by bioenergy expansion (1.03 times globally and 1.23 times over land), primarily driven by reduced total precipitable water and relative humidity, as revealed by an interpretable machine learning framework, which identifies biophysical warming of the boundary layer due to albedo-driven surface heating and enhanced sensible heat flux as the dominant mechanism. Forest darkening reduces surface albedo, which increases absorbed solar radiation and elevates sensible heat, thereby drying the boundary layer and suppressing cloud formation. Conversely, realistic afforestation  mitigates the loss of low- and mid-level clouds and suppresses boreal warming. Our findings indicate that more forest expansion does not always generate greater climate benefits; the climatic outcome largely depends on the type of land conversion and specific latitude bands, and highlights the critical importance of carefully selecting afforestation areas in the future to achieve positive climate benefits.