The Contours of the Cloud: Dissecting the Real Estate Investment Decisions of Data Center Operators

Abstract

This thesis investigates the real estate investment decisions of data center operators, with a focus on how key infrastructure characteristics influence data center development. Using a sequential econometric approach, the research applies both a logit and a hedonic model to evaluate the importance of various factors. The logit model explores the likelihood of data center development at the county level, highlighting geographical characteristics. The hedonic model examines the impact of specific site attributes, such as proximity to power infrastructure and fiber, on the scale of data center facilities in megawatts. The findings suggest that colocation data centers prioritize connectivity, electrical infrastructure, and urban proximity, while the location of hyperscale facilities is more variable and less predictable. This study enhances our understanding of how modern technological demands, particularly in the AI era, shape real estate strategies and offers insights into future trends in digital infrastructure investments.