The role of cloud feedback in Arctic amplification (AA) of anthropogenic warming remains unclear. Traditional feedback analysis diagnoses the net cloud feedback as positive in the tropics but negative in the Arctic, suggesting that AA would be amplified if cloud feedback were suppressed. However, in cloud-locking experiments using the slab ocean version of Energy Exascale Earth System Model (E3SM), we find that suppressing cloud feedback results in a substantial decrease in Arctic amplification under greenhouse gas forcing. We find that the increase in AA from cloud feedback arises from three mechanisms: 1) cloud feedbacks trap more heat in the tropics, leading to increased poleward moist atmospheric heat transport (AHT) which then amplifies Arctic warming; 2) the additional Arctic warming increases the warming impact of positive non-cloud feedbacks in the region (such as the surface albedo feedback); and 3) changes in warming structure due to cloud feedbacks modify the strength of local non-cloud feedbacks, making the water vapor feedback more positive in the tropics but the surface albedo feedback less positive in the Arctic. We further evaluate the relative contributions of these three processes using a moist energy balance model (MEBM), which demonstrates that interactions of cloud feedback with AHT and with other feedbacks – the processes described in (1) and (2) – dominate the total effect of the interactive cloud. It also shows that while changes in local non-cloud feedbacks increase global warming by about 11% (and Arctic warming by 9%) when cloud feedbacks are suppressed, their contribution to Arctic amplification is relatively small. These results demonstrate that traditional attributions of AA, based on local feedback analysis, may be misleading, as they overlook key interactions between non-local cloud changes, poleward AHT, and non-cloud feedbacks in the Arctic.