<p>Energy-intensive industries closely depend on fossil fuels and contribute substantially to carbon emissions in China. Examining the carbon emission performance and its influencing factors in these industries is imperative for sustainability. Firstly, we assess carbon emission efficiency and Luenberger carbon productivity with the nonradial directional distance function. The spatial characteristics are revealed by the exploratory spatial analysis. Secondly, we detail the factor contribution to Luenberger carbon productivity and the decomposed indicators. The evolutions of carbon emission efficiency are displayed by kernel density analysis. Finally, we examine the influencing factors of carbon emission efficiency with the spatial Durbin model from the perspective of transition and upgrading. The main insights indicate that (1) the efficiency values decrease from southeastern China to western China, and they display a centripetal clustering in the northeast-southwest direction. (2) The over-expansion industrial scale contributes to the decrease in carbon productivity, and the carbon emission reduction and the total input saving promote its increase. The labor saving drives the increase in carbon productivity, whereas energy consumption initially drives its growth but finally contributes to the decline. The driver of the transition has shifted from efficiency changes to technical changes. (3) Industrial agglomeration shows an inverted U-shaped impact on efficiency, and energy structure displays a negative impact on it. Industrial upgrading and green innovation show positive impacts on efficiency. Industrial agglomeration, industrial upgrading, and green innovation reflect spatial spillover and siphoning effects. The findings have important policy implications for improving sustainability of energy-intensive industries.</p>