This article compares the robustness of the optimal choice of technologies for two Smart Energy Systems architectures at district level, illustrated by a case study representative of a newly built district in Grenoble, France. The electricity-driven architecture relies on the national electric grid, decentralized photovoltaic panels and decentralized heat pumps for heat production building by building. The alternative architecture consists of a district heating network with multiple sources and equipment for centralized production of heat. Those are a gas boiler plant, a biomass-driven cogeneration plant, a solar thermal collector field, and a geothermal heat pumping plant (grid-driven or photovoltaics-driven). Electric and heat storages are considered in both architectures. The sizing and operation of both architectures are optimized via linear programming, through a multi-objective approach (total project cost versus carbon dioxide emissions). Both architectures are compared at nominal scenario and at sensitivity scenarios. It is concluded that the electricity-driven architecture is less robust, especially to uncertainties in space heating demands (+150%/−30% impact on costs) and in heat pump performance (+35%/−15% in costs). Meanwhile, the multi-source architecture is less sensitive to space heating demands (+110%/−30%) and has negligible sensitivity to the rest of parameters except photovoltaic panels efficiency (+14%/−7%).