The Tibetan Plateau (TP) is a global warming hotspot, however, the warming status at high elevation (>5000 m) is poorly understood due to very sparse observations. Here we analyze spatial patterns in TP warming rates based on a novel near-surface air temperature dataset of 1980–2014 recently developed by ingesting high-elevation observations and downscaled reanalysis datasets. We show that the high snow cover persistence at high elevation reduces strengthening of positive feedbacks responsible for elevation dependent warming at low-middle elevations, leading to reversed altitudinal patterns of TP warming above and below 5000 m. An important negative feedback is induced by the presence of snow and glaciers at elevations above 5000 m, due to their “buffering” effects by consuming or reflecting energy that would be used for warming in the absence of snow or ice. A further decrease in snow cover and glacier extent at high elevations may thus amplify the warming on the TP.