Résumé : The systems that operate the infrastructure of cities have evolved in a fragmented fashion across several generations of technology, causing city utilities and services to operate sub-optimally and limiting the creation of new value-added services and restrict opportunities for cost-saving. The integration of cross-domain city data offers a new wave of opportunities to mitigate some of these impacts and enables city systems to draw effectively on interoperable data that will be used to deliver smarter cities. Despite the considerable potential of city data, current smart cities initiatives have mainly addressed the problem of data management from a technology perspective, and have disregarded stakeholders and data needs. As a consequence, such initiatives are susceptible to failure from inadequate stakeholder input, requirements neglecting, and information fragmentation and overload. They are also likely to be limited in terms of both scalability and future proofing against technological, commercial and legislative change. This paper proposes a systematic business-modeldriven framework to guide the design of large and highly interconnected data infrastructures which are provided and supported by multiple stakeholders. The framework is used to model, elicit and reason about the requirements of the service, technology, organization, value, and governance aspects of smart cities. The requirements serve as an input to a closed-loop supply chain model, which is designed and managed to explicitly consider the activities and processes that enables the stakeholders of smart cities to efficiently leverage their collective knowledge. We demonstrate how our approach can be used to design data infrastructures by examining a series of exemplary scenarios and by demonstrating how our approach handles the holistic design of a data infrastructure and informs the decision making process.