Industrial companies already apply digital twins for the digital representation of the physical world. In addition, information sharing becomes increasingly decisive for the competition, e.g. in supply networks where products and logistics data such as demands and capacities are exchanged. As many companies are, however, highly reluctant to share data across the supply chain, this paper applies the methodology of design science research to, first, state the requirements for shared digital twins based on five industrial use cases. It turns out that with regard to data decentralism, sovereignty and compatibility through global standardization are key success factors. Hence, second, this paper presents a concept for a shared digital twin providing data on demand, i.e. at the right time and in particular with data condensed to the concrete need.