We estimate the development effects of electrification across Brazil over the period 1960-2000. Brazil relies almost exclusively on hydropower, which requires intercepting water at high velocity. We build an engineering model which takes as inputs only geography (river gradient, water flow and Amazon) and simulates a time series of hypothetical electricity grids for Brazil that show how the grid would have evolved had infrastructure investments been made based solely on geologic cost considerations, ignoring all demand-side concerns. Using the model as an instrument, we document large positive effects of electrification on development that are underestimated when one fails to account for the political allocation of infrastructure projects or its targeting to under-developed areas. Broad-based improvement in labor productivity across sectors and areas rather than general equilibrium re-sorting (in-migration to electrified counties) appears to be the likely mechanism by which these development gains are realized.