According to the EU policy, a future-proof broadband supply for all European households is to be achieved by 2025. There is already a wide range of fibre deployment in Europe. However, the expansion of fibre-based access networks in Europe to date has taken place mainly in large cities. In other areas, the expansion is sluggish or non-existent. As a result, a digital divide between urban and rural areas in Europe is arising. The spatial disparity in fibre roll-out is often justified by market stake holders with significant regional cost differences. In the absence of private-sector investment, government subsidy programmes are often used to improve broadband coverage. Thus, politicians have to deal with the question about the level of investment required and the spatial distribution of subsidy needs. In this paper, we will therefore investigate the question of how significant the heterogeneity in the costs of Very High Capacity (VHC) networks in Germany actually is and whether and how the costs for Very High Capacity (VHC) networks differ between urban and rural regions. In the first part of the paper, we will analyse the regional cost differences of access network areas on the basis of bottom-up calculated investment figures. In the second part of the paper, we establish statistical estimation models that explain these regional cost differences. For this purpose, we use publicly available data. As a reference value for regionally differentiated costs of Very High Capacity (VHC) access networks, we use the results of a detailed bottom-up modelling of an FTTH network carried out for the whole of Germany. The model uses georeferenced household and business location data and optimizes the access network routes along the street network in a bottom-up manner. This model allows us to determine regionally differentiated FTTH investment at the level of access areas. By matching this data with the EU-wide standardized EUROSTAT urban/rural typology classification (predominantly urban, intermediate and predominantly rural), we determine whether and to which extent significant regional cost differences can be found in Germany applying these classifications. One focus is on determining the spread of investment requirements, especially among rural areas. Based on our experience, these areas exhibit the lowest economic viability of a network roll-out and, thus, the highest need for funding. By using statistical indicators, we analyse the suitability of the EUROSTAT classification as a differentiation criterion for regional cost differences. Here, we are particularly interested in whether the areas defined as rural form a sufficiently homogeneous group, and whether they show comparable levels of required investment. Our findings confirm that the differentiation criterion used, namely EUROSTAT urban/rural typology classification, is not satisfactory in measuring regional cost differences. It cannot sufficiently account for a large share of observable differences in fibre-based access network costs. Since it is desirable to answer questions regarding the required funding for selected regions based on publicly available data, we apply regression models to identify alternative influencing factors on the basis of publicly available data, in order to better explain observable regional cost differences. Here, we find that a handful of geographical factors are capable of explaining 95% of the geographical differences in fibre investment requirements, the most relevant being the number of connection lines, the number of households per kilometre of road in built-up areas, the main road length per built-up area and the share of built-up area in relation to overall area. In the last part of the analysis, we examine whether the derived results are also meaningful in a political and regulatory context. Discussions about the necessity of promoting high-speed networks usually take place at the level of local authorities. Therefore, in a final step, we address the question whether the statistical relationships derived from the regression model at the level of access areas also apply at a higher aggregated, i.e. NUTS3, level. In summary, we show that for Germany, classifications based on subscriber density exhibit a significant spread in the investment costs of Very High Capacity (VHC) access networks, which is most pronounced in rural clusters. Statistical analyses using regression models can improve the result if geographical elements of the settlement structure are considered in the analysis.