Reconciling Global Data-Centre Electricity Projections for 2030: A Boundary-Aware Evidence Map
Abstract: Background: Published projections of global data-centre electricity use are frequently compared despite differences in geography, facility and workload coverage, units, and scenario design. Methods: We constructed a secondary-source-seeded, boundary-aware evidence map of projections for annual global operational electricity use by total data centres in 2030. Each publication was coded across eight comparability dimensions and assigned a transparent primary-source verification status. The primary synthesis treats publications reporting only range endpoints as interval-censored rather than interpreting their endpoints as complete scenarios. Publication-level bootstrap intervals and verification, peer-review, scenario-completeness, aggregation, and influence sensitivities were examined. Results: The corpus contains 28 values from 15 publications; 7 were directly checked in primary or author full text by one reviewer and 8 remain verification incomplete. The interval-censored model estimates a median latent publication centre of 912 TWh yr-1 (95% publication-bootstrap interval 712-1217) and a geometric standard deviation of 1.69. A conventional geometric-scenario summary gives a median of 945 TWh yr-1 (interquartile range 648-1167). Method, publisher, and vintage contrasts are imprecise and are retained only as exploratory sensitivities. Conclusions: Boundary harmonisation makes the literature more interpretable but does not identify a probability distribution for realised 2030 demand. The principal contribution is an auditable comparability taxonomy and minimum reporting standard, not a consensus forecast.
Keywords: data centres; electricity demand; forecasts; evidence mapping; interval censoring; system boundaries; reporting standards
How to Cite?: Aditya Singh, "Reconciling Global Data-Centre Electricity Projections for 2030: A Boundary-Aware Evidence Map", Volume 15 Issue 10, October 2026, International Journal of Science and Research (IJSR), Pages: 664-672, https://www.ijsr.net/getabstract.php?paperid=SR26901133612, DOI: https://dx.doi.org/10.21275/SR26901133612