Food production today is more global than ever. Food trade ensures adequate and diverse food even in areas with low self-sufficiency. Foodstuffs are being traded across the world, but so are agricultural inputs such as fertilizers, machinery, and pesticides. Shocks and disturbances in the trade flows of agricultural inputs, caused by e.g. conflict, may be devastating to the food production and yields of otherwise self-sufficient countries. This aspect of food security and resilience requires more attention. In this study, we modelled the effects of agricultural input shocks using global spatial data on crop yields, fertilizers, machinery and pesticides with random forest, a machine learning algorithm. We show that the most drastic yield losses are caused by shocks in one or multiple fertilizers. Areas with the highest crop yields suffer the most from all agricultural input shocks, while low-yielding areas are seldom affected. Yield losses in these high-yielding ‘breadbasket’ areas of the world would be detrimental to global food security. Our study provides important information in high spatial definition to be used in future discussions on food security and resilience.