Phosphorus (P) is a finite resource and critical to plant growth and therefore food security. Regional- and continental-scale studies propose how much P would be required to feed the world by 2050. These indicate that sub-Saharan Africa soils have the highest soil P deficit globally. However, the spatial heterogeneity of the P deficit caused by heterogeneous soil chemistry in the continental scale has never been addressed. We provide a combination of a broadly adopted P-sorption model that is integrated into a highly influential, large-scale soil phosphorus cycling model. As a result, we show significant differences between the model outputs in both the soil-P concentrations and total P required to produce future crops for the same predicted scenarios. These results indicate the importance of soil chemistry for soil-nutrient modelling and highlight that previous influential studies may have overestimated P required. This is particularly the case in Somalia where conventional modelling predicts twice as much P required to 2050 as our new proposed model.