In the plenary talk I will discuss the causes of higher climate sensitivity in the latest state-of-the- art CMIP6 Earth system models relative to their predecessors in CMIP5. Taken as a whole, these latest models are more sensitive primarily because they have stronger amplifying cloud feedbacks, particularly in the extratropics, where low clouds become less extensive and less reflective as the planet warms. This stronger positive cloud feedback arises due to changes in model physics and may be related to improved representation of cloud phase. Given the continued importance of low cloud feedback in driving uncertainty in climate sensitivity across climate models and between model generations, I will then discuss ongoing work that is attempting to constrain the global marine low cloud feedback using satellite observations of how low cloud properties respond to individual cloud-controlling factors. This work indicates that the observed sensitivity of low clouds to their meteorological controls is incompatible with very high or very low values of climate sensitivity