Mapping the impact of artificial intelligence on energy poverty: New evidence from spatial panel models

Energy poverty remains a critical challenge for sustainable development, particularly in low- and middle-income countries. As countries seek innovative solutions to expand energy access, artificial intelligence (AI) has emerged as a promising tool. While recent studies have explored the role of AI in improving energy access, few have considered its spatial effects. Therefore, this paper investigates how AI adoption affects energy poverty using a spatial panel dataset of 64 countries from 2010 to 2019. Spatial econometric models reveal that higher AI adoption is significantly associated with reductions in energy poverty and that these benefits extend beyond national borders through regional spillovers. Mediation analysis shows that technological innovation, proxied by patent activity, partially transmits the impact of AI, while moderation analysis reveals that the effect of AI is stronger in less urbanized settings and where public spending is relatively low. These findings provide the first empirical evidence of spatial dependence in the AI–energy poverty nexus and highlight the importance of designing targeted, regionally coordinated policies. Thus, promoting AI-enabled off-grid solutions and strengthening innovation systems could help reduce spatial disparities in energy access, especially when embedded within broader international partnerships and adaptive national energy policies.