Local Computing vs. Cloud Computing: An Empirical Study of Energy Consumption

The dominance of Software-as-a-Service (SaaS) has raised sustainability concerns due to its reliance on centralized cloud infrastructure and increasing energy consumption. This paper investigates a core principle of the Local-First Software (LoFi) paradigm, local processing (LP), as a potential alternative. Through a controlled empirical study using SeekTune, an open-source music recognition application, we compare the end-user device and server energy consumption and data usage of an LP architecture against a traditional SaaS model for a computationally intensive task. Our results demonstrate that, while LP consumes more energy on the client device, the combined energy cost of data transmission and server-side computation in the SaaS model often results in a larger total systemic footprint. In addition, LP transmits considerably less data over the network than SaaS. For modern mobile devices, processing locally can be significantly more energy-efficient overall. These findings challenge the long-held assumption that cloud offloading is inherently superior, and position the LP aspect of LoFi as a crucial consideration for sustainable software design.