Existing work on the modernization theory has focused on the effect of GDP per capita on democracy. This implicitly assumes that a 1% increase in average per capita income from agricultural production by big landowners, resource extraction, tax evasion, or tourism, is equally likely to lead to democracy as a 1% increase in income from manufacturing or modern services. I question this assumption and argue that it is primarily industrialization, and not economic growth in general, that induces democracy. In line with this hypothesis I find that industrialization is strongly correlated with democracy across 141 countries over 170 years (1845-2015), and that all existing challenges to modernization theory disappear when one focuses on industrialization, rather than GDP per capita -- e.g., the effect is robust to country and time fixed effects, occurs on both transitions and consolidations, and is present in all time periods. Two novel instruments suggest that the effect is causal.