Smartphones, personalized from applications to notifications, are now a dominant source of political information, yet little is known about the content people consume on them. We provide the first systematic analysis of the magnitude and drivers of election-related smartphone consumption, using novel data on the content people observed on their phones during the 2024 U.S. election campaign. Despite a highly contentious campaign, the median American consumed limited election content on their phones, and consumption remained stable over time. Consumption primarily came from applications with personalized content (social media and music & videos apps), with less than 10% from dedicated news applications. To understand the drivers of this pattern, we examine heterogeneity across both applications and individuals. We find that across applications, individuals consumed more election content on X and Reddit than on Facebook or Instagram. Across individuals, consumption was positively skewed, with higher consumption among swing-state residents, news app users, and even fans of a non-political figure (Taylor Swift) following her endorsement. Using a variance decomposition exercise, we find that differences across individuals, rather than differences between the applications they use, drive the observed heterogeneity. These findings suggest that policies altering voter incentives to consume news -- rather than regulating platform algorithms -- may be more effective for increasing political engagement.