In many cities, pilot projects are set up to test or develop new technologies that improve sustainability, urban quality of life or urban services (often labelled as “smart city” projects). Typically, these projects are supported by the municipality, funded by subsidies, and run in partnerships. Many projects however die after the pilot stage, and never scale up. Policymakers on all levels consider this as a challenge and search for solutions. In this paper, we analyse the process of upscaling, focusing on smart city projects in which several partners –with different missions, agenda’s and incentives- join up. First, we review the extant literature on upscaling from development studies, business studies, and the transition management literature. Based on insights from these literatures, we identify three types of upscaling: roll-out, expansion and replication, each with their own dynamics, context sensitivity and scaling barriers. We illustrate the typology with recent smart city projects in Amsterdam. Based on desk research and in-depth interviews with a number of project stakeholders and partners of the Amsterdam Smart City platform, we analyse three projects in depth, in order to illustrate the challenges of different upscaling types. i) Energy Atlas, an EU-funded open data project in which the grid company, utilities and local government set up a detailed online platform showing real-time energy use on the level of the building block; ii) Climate Street, a project that intended to make an entire urban high street sustainable, involving a large number of stakeholders, and iii) Ikringloop, an application that helps to recycle or to re-use waste. Each of the projects faced great complexities in the upscaling process, albeit to a varying degree. The paper ends with conclusions and recommendations on pilot projects and partnership governance, and adds new reflections to the debates on upscaling.