Based on the panel data from 2009 to 2019, this paper used the carbon emission coefficient method to calculate the total carbon emissions of the four national-level urban agglomerations and the carbon emissions of fossil energy, secondary energy and building materials. The results show that there are significant spatial differences in the total carbon emissions of the four national urban agglomerations. The peak value of the Yangtze River Delta urban agglomeration was 967 million tons in 2011, and that of the Guanzhong Plain urban agglomeration was 88 million tons in 2019. Based on the calculation results of energy carbon emissions, the LMDI model was further used to decompose the influencing factors into five major effects, including carbon emission coefficient, energy intensity, energy structure, economic level of construction industry and population size. It is found that the carbon emission coefficient has the least effect on the carbon emission of energy, the economic level of construction industry plays a promoting role, and the energy structure plays a restraining role. In the future, it is of significance to promote the use of new building materials, optimize the energy structure of the construction industry and promote low-carbon and energy-saving technology in the construction industry to promote carbon reduction.