Wireless sensor networks (WSNs) are attracting attention in the context of smart grids as they can facilitate the development of future energy consumption monitoring systems. In this paper, we present a comparison of the performance of four WSN architectures for energy consumption monitoring. The aim of this work is to explore the benefits of having distributed storage and processing compared with centralized approaches. Performance of the architectures was evaluated, considering energy consumption, processing, storage and bandwidth requirements, as well as capital costs involved. Results show that architectures requiring distributed processing and storage significantly improve the energy efficiency; nonetheless, deploying these architectures requires a relatively higher investment in comparison to centralized approaches.