Extratropical cyclones (ETCs) are the primary drivers of severe weather over the North Atlantic, yet projections of changes in the intensity of the most extreme storms remain highly uncertain. This study investigates inter-model uncertainty in future climate projections of extreme cyclones in winter, arising from competing processes such as reduced midlatitude baroclinicity and enhanced moisture availability. We analyze the future changes in the most intense 100 ETCs across 13 models from the Coupled Model Intercomparison Project Phase 6 (CMIP6) under the highest forcing scenario (SSP5-8.5). Results show a robust increase in 850 hPa relative vorticity found in 11 of 13 models, signaling an intensification of extreme cyclones despite reduced baroclinicity (9/13 models). Precipitation associated with extreme cyclones intensifies (11/13 models) despite near surface temperature gradient weakening, with results suggesting that the weakening of dry baroclinic processes is offset by enhanced diabatic feedbacks. Surface winds show no consensus, highlighting the large variability across models in predicting cyclone-related winds. Spatially, extreme storms in the North Atlantic tend to exhibit an eastward shift towards Europe (7/13 models), with mixed latitudinal responses. Overall, these findings show a projected intensification of extreme midlatitude cyclones in the North Atlantic, accompanied by robust thermodynamic signals related to an intensified precipitation in most models. Further assessments utilising large ensembles of climate models, alongside systematic investigations of extreme cyclone energetics can help to further clarify the relative importance of dry baroclinic versus moist diabatic processes.