This research examines the Environmental Kuznets Curve (EKC) hypothesis in G20 countries over 58 years, focusing on CO2 emissions. Using the Auto-Regressive Distributed Lag (ARDL) model with Feasible Generalized Least Squares (FGLS), the analysis accounts for heteroscedasticity in the observations. The findings reveal that industrial value-added per capita increases long-term emissions, whereas agricultural value-added per capita helps reduce emissions due to its environmentally friendly nature. Energy consumption significantly contributes to emissions, while urbanization exacerbates emissions in developing countries but tends to mitigate them in developed nations. A key insight is that GDP per capita in developing countries negatively impacts CO2 emissions in both the short and long term. In contrast, developed countries initially experienced rising emissions during industrialization but later reduced them by shifting to a service-based economy. It refers to the term "Dual-Path Emission Development", suggesting that developing countries do not necessarily have to follow the traditional EKC pattern of rising emissions before achieving reductions. This highlights the need for differentiated policies between developed and developing nations in addressing CO2 emissions.JEL: Q50, C22.