Current electricity grids do not fit the needs and challenges of the 21st century, such as the need to transition to renewable energy sources and the variability in power supply concomitant with such energy sources. In this context, smart electricity grids have been proposed as a solution. A large number of pilots and experiments have been set up, but a key challenge remains how to upscale them. Upscaling is critically important to enable a wide scale integration of renewable energy sources. This paper mobilises literature on the strategic management of experimental niches to explore the upscaling of smart grids in the Netherlands. On the basis of existing literature, a typology of four different patterns of upscaling is proposed: growing, replication, accumulation, and transformation. The relevance of this typology to understanding upscaling of smart grids is explored in a comparative qualitative case study design. On this basis we argue that the building of broad and deep social networks is important for growing and replication; articulating and sharing expectations is important for replication; and broad and reflexive learning processes are critical to transformation and replication. The paper concludes by arguing that these findings can provide important guidelines for future energy innovation policies.