Get started
Work

Carbon Emission in Cloud Data Centers: A Comparative Analysis of Cloud and On-Premise Infrastructure

Year: 2026
Type: preprint
Abstract: Cloud computing has transformed the delivery of digital services; however, its rapid expansionhas resulted in a substantial increase in global energy consumption and associated carbonemissions. This study examines the environmental sustainability of cloud infrastructure through asystematic review of secondary data on data center energy metrics and the primary factorscontributing to carbon footprints. In addition, a comparative analytical model is developed toevaluate the energy efficiency and carbon impact of cloud-based systems relative to traditionalon-premise infrastructure under three grid carbon intensity scenarios. The results demonstratethat cloud platforms achieve approximately 87.8% lower annual energy consumption perequivalent workload compared to traditional on-premise data centers under identical gridconditions, primarily due to superior Power Usage Effectiveness (PUE) of 1.1 versus 1.8 andhigher server utilization rates of 65% versus 15%. This operational efficiency translates directlyto an 87.8% reduction in carbon emissions when grid conditions are held constant; when cloudinfrastructure is powered by a renewable energy grid (CI = 50 gCO₂e/kWh), emissions decreaseby a further 89.5% relative to fossil-grid cloud deployment, reaching 1,446 kg CO₂e annuallycompared to 112,347 kg for on-premise systems. Despite these efficiencies, the continuedgrowth of cloud services driven by AI workloads and big data presents ongoing sustainabilitychallenges. This study concludes that cloud adoption is a viable and significant pathway forreducing organizational carbon footprints, provided that renewable energy integration andoperational efficiency improvements keep pace with rising global demand.
DOI: 10.5281/zenodo.22031852
Language: English
Open Access status: green
OpenAlex ID: W7203940806
Cites:
Cited by:
Related to: