While public debate and prominent studies expect digitalization to substantially reduce energy use and carbon dioxide (CO2) emissions, quantitative research has produced ambiguous results. This study addresses the challenges in the analysis of the relationship between a country’s digitalization level and CO2 emissions by employing the Group Fixed Effects estimator for panel data of EU and OECD countries and by differentiating between emissions associated with digitalization in firms and households. Results are highly robust to the statistical procedure and indicate that digitalization in both households and firms generally decreases emissions. At the sample median, a 10% increase in firm (household) level digitalization would, on average, decrease emissions by 0.3% (0.8%). In countries of the three lowest deciles in the income distribution, however, the effect is reversed: Here, an increase in digitalization is also associated with an increase in emissions. The results are further interpreted beyond the median effect and differentiate between countries of different incomes through a non-parametric approach. This analysis also has implications for the discussion of the Environmental Kuznets Curve (EKC) hypothesis, as the empirical analysis nests an estimation of an EKC model, extended by a measure of digitalization.