Artificial intelligence and global embodied carbon flow: Evidence from the application of industrial robots

The rapid advancement of artificial intelligence (AI), exemplified by industrial robots, holds significant potential for reducing global carbon emissions. However, its impact on global trade-embodied carbon flows remains ambiguous. Using instrumental variable methodology, this study empirically examines how industrial robot adoption in developing countries affects embodied carbon exports. The key findings are as follows: First, increased industrial robot usage in developing countries significantly boosts their embodied carbon outflows, with a 1-unit rise in robot adoption leading to an additional 0.262 thousand tons of embodied carbon exports. This effect is particularly pronounced in manufacturing and utilities, amplifying the risk of carbon leakage to other regions. Second, industrial robots disproportionately drive embodied carbon flows among developing nations. From an industry perspective, robot adoption primarily increases carbon outflows in medium-carbon-intensity sectors. Mechanism analysis reveals that these effects stem from three main channels: expanded production and trade scales, higher energy consumption, and reduced reliance on renewable energy. These findings challenge conventional wisdom by highlighting two critical risks: automation may inadvertently create new carbon leakage pathways within the developing world, and technological leapfrogging without complementary energy reforms could exacerbate climate inequities.