CO2 emissions are the pressing issue of our time, burdening humanity. To reduce CO2 emissions and global warming, there is only one solution: sustainability and the expansion of renewable energies. However, it does not help if one part of the world reduces emissions while another increases CO2 emissions. This task must be solved together. This paper addresses modern machine learning algorithms that utilize the available data from all countries from 2000 to 2020. The extensive data ranges from access to electricity to geographical and economic figures to make country-specific CO2 predictions. In this paper, we use various machine learning approaches to predict CO2 emissions and evaluate these approaches in terms of accuracy and predictive suitability. The focus will be on the algorithms and applied techniques that lead to the best predictions.