Green Cloud? An Empirical Analysis of Cloud Computing and Energy Efficiency
by
Kunsoo Han
Associate Professor
Bensadoun Faculty Scholar
Desautels Faculty of Management
McGill University
Friday, Mar 26
9 – 10 am | Zoom
(send an email to ayabakan@temple.edu to get the Zoom link)
Abstact:
The rapid, widespread adoption of cloud computing over the past decade has sparked debates on its environmental impacts among practitioners and academics alike. While prior literature has extensively scrutinized the energy-consuming characteristics of cloud computing on the vendor side, very little attention has been paid to its broader, economy-wide impacts on the client side, although the latter effect may be substantially greater. Drawing upon production theory and using a stochastic frontier analysis, this study examines the impact of cloud computing on client industries’ energy efficiency. To this end, we develop a novel industry-level measure of cloud-based information technology (IT) services. Using U.S. economy-wide data from 57 industries during 1997–2017, our findings suggest that cloud-based IT services improve energy efficiency, and this effect appears to be driven mainly by Software-as-a-Service (SaaS), rather than Infrastructure-as-a-Service (IaaS). Moreover, this effect is significant only after 2006, when cloud computing started to be commercialized, and becomes even stronger after 2010. While SaaS contributes to energy efficiency improvement across industries, we find that the positive contribution of IaaS to energy efficiency is significant only in IT hardware-intensive industries. Finally, the main effects of SaaS on energy efficiency are universally significant for both electric and non-electric energy efficiency, whereas IaaS is positively associated with electric energy efficiency when IT hardware intensity is high. From the analysis, we estimate the economy-wide savings in energy costs due to cloud computing to be 2.8 to 12.6 billion US dollars in 2017 alone, equivalent to a reduction in electricity use by 31.8 to 143.8 billion kilowatt-hours. Our estimate exceeds the total energy expenditure in the cloud services vendor industries and is comparable to total electricity consumption in U.S. data centers. Additional firm-level survey analysis corroborates the findings from industry-level econometric analysis. We discuss implications for research and practice.