This study proposes a new approach to determine sustainable agro-industry logistics by minimizing supply and demand imbalances using spatial analysis. The potato food crop agro-industry at Wonosobo district, Central Java, Indonesia, was selected as the case. This study first developed a spatial-based cropland sustainability classification. Second, developed a sustainable harvest prediction by modifying the multi-thresholding using remote sensing and IoT, namely SHT15 and rain gauge sensor. Third, we developed a spatial route planning wherein we adjust the COG and spatial Dijkstra algorithm to select the optimal route. This study developed a comprehensive sustainable analysis by integrating spatial dimensions with environmental, economic, and social dimensions to support sustainable food security. Multi-criteria spatial analysis was used in an environmental measurement considering altitude, soil texture, slope, rainfall, humidity, and temperature. In the economic dimension, we compare the predicted total harvests with the current production. We consider the social dimension, namely population density, spatial-temporal congestion, and risk hazard-zones index. This study shows that spatial-based cropland sustainability classification can determine the most sustainable location for potato plants. This study shows that the sustainable harvest prediction by the modified multi-thresholding can predict the harvests more accurately, namely 89.3%. Spatial distributing route planning with adjustment of the COG obtains more rational coordinates than the classical method. In addition, spatial Dijkstra algorithm modification can show a more optimal route for agro-industrial commodities. This new approach has demonstrated that spatial analysis can be an alternative solution to minimize supply and demand imbalances in sustainable agro-industry logistics.