Today, more than 4 billion people face water shortages at least once a year. Water scarcity is projected to aggravate due to the pressure that climate change and increasing population exert on freshwater resources. Yet, projections of water scarcity remain highly uncertain and vary with climatic conditions around the world. Risk assessments of water scarcity remain focused on the local balance between water supply through precipitation and human water demand, and thus do not consider spatio-temporal dependencies of supply and demand. However, over the continental land, around 40% of precipitation originates from terrestrial evaporation, often coming from neighbor regions or even remote locations. Here, I propose to assess the risk of water scarcity worldwide by investigating the origin of precipitation. Using the traditional notion of watersheds, I aim to establish a global precipitation recycling network that quantifies the atmospheric flow of water from evaporation to precipitation. An integrative risk assessment of water scarcity will be enabled by (i) highlighting watersheds that are strongly dependent on the water supply from other watersheds in the network, (ii) unraveling the propagation of drought through the network, and (iii) quantifying the vulnerability of each watershed to remote drought. This risk assessment will establish an atmospheric perspective on water scarcity, which may ultimately help develop adaptation strategies to secure water availability.