Digital financial inclusion (DFI) is newly cited as a significant driver of sustainable development in developing and transition economies, and it is thus likely that DFI affects energy-environment performance (EEP). Here, different from recent studies documenting monotonic links between them, we first theoretically reconcile the underlying U-shaped impact of DFI on EEP using two competing hypotheses: consumption stimulation versus innovation promotion effect. Subsequently, we extend non-radial direction distance function specifications within a biennial data envelopment analysis framework to assess EEP, and then re-examine the causal relationship between DFI and EEP using panel data covering 282 prefecture-level cities in China during 2011-2019. The empirical results show that DFI has a solid U-shaped impact on the EEP instead of a linear one. This effect holds up to various robustness and endogeneity checks, including the spatial econometric approach that captures cross-sectional dependence, the geographical distance as an instrument variable, the dynamic threshold regression approach, and the difference-in-difference approach. Further mechanism tests show that household consumption stimulation and green innovation promotion are two plausible economic channels to explain this U-effect, supporting our theoretical conjecture. Moreover, the U-effect is more pronounced for cities with relatively low levels of economic development and environmental regulation, including central and western cities, as well as smaller, non-low-carbon pilot and non-carbon emission trading pilot cities. However, we find that the U-effect is disproportionately stronger in southern cities with higher financial marketization levels than in northern cities. Overall, these findings shed more light on the DFI-EEP nexus and provide new insights into understanding the real effects of DFI on sustainable development.