Characterizing the Usage Intensity of Public Cloud

Abstract

This article uses precise and novel data on country-level Cloud IaaS and PaaS revenue to measure the intensive margin of technology diffusion across countries and within countries over time. We horse race diffusion models and find that cloud diffusion exhibits both Log-Log and Logistic Growth patterns. We use cross validation on nearly 100 features to determine what correlates with cross-country differences. We find that increases in features impacting Gross Domestic Product, Internet Connectivity, and Human Capital are associated with increases in intensity of cloud adoption. We finally compare the relative impacts of these variables using a random coefficients model. Although correlative, our algorithmic research design motivates data-driven hypothesis generation and further causal work regarding how policymakers can encourage more cloud computing adoption and technology adoption more broadly.