Bridging FinOps practice and machine learning research: a systematic review of cloud cost optimization
PeerJ Computer Science · 22.06.2026 · 12 · e3964
DOI: 10.7717/peerj-cs.3964 ↗Kurz gefasst
Systematische Übersicht in der Zeitschrift PeerJ Computer Science, online erschienen am 22.06.2026. Verfasst von Mateen Ali Anjum. Die Arbeit nennt 79 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
Cloud computing expenditures continue to climb, with organizations reporting 30–35% of cloud spend as waste. FinOps (Cloud Financial Operations) has emerged as a discipline to manage cloud costs, yet academic research on machine learning approaches for FinOps remains fragmented and disconnected from practitioner frameworks.
Objectives
This systematic review bridges the gap between FinOps practice and machine learning (ML) research. It maps ML techniques to the FinOps Foundation’s capability framework, analyzes the growing role of Large Language Models (LLMs) in cloud cost management, evaluates how studies report (or fail to report) business-oriented metrics, and identifies organizational barriers to ML adoption.
Methods
Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, searches were conducted across IEEE Xplore, ACM Digital Library, Springer, ScienceDirect, arXiv, and grey literature sources (2019–2025). Peer-reviewed articles and substantial industry reports addressing machine learning for cloud cost optimization, prediction, or anomaly detection were included.
Results
The review analyzed 75 academic studies and four grey literature sources (79 total). The resulting taxonomy links six ML technique categories to FinOps capabilities across the Inform, Optimize, and Operate phases. Among the 17 LLM-related articles, applications range from inference cost optimization to autonomous agents. Business metrics remain scarce: only 19% of studies report cost outcomes. Organizational barriers, notably skill gaps (65% of organizations), data quality issues, and cultural resistance, consistently impede adoption.
Conclusions
ML techniques show promise for FinOps automation, but wide gaps persist between academic research and practitioner needs. A research agenda addressing LLM integration, business metric standardization, and organizational adoption is proposed to advance the field.
Zitieren
APA
Anjum, M. A. (2026). Bridging FinOps practice and machine learning research: a systematic review of cloud cost optimization. PeerJ Computer Science, 12, e3964. https://doi.org/10.7717/peerj-cs.3964
Vancouver
Anjum MA. Bridging FinOps practice and machine learning research: a systematic review of cloud cost optimization. PeerJ Computer Science. 2026;12:e3964. doi:10.7717/peerj-cs.3964
BibTeX
@article{Anjum2026,
author = {Anjum, Mateen Ali},
title = {Bridging FinOps practice and machine learning research: a systematic review of cloud cost optimization},
journal = {PeerJ Computer Science},
year = {2026},
volume = {12},
pages = {e3964},
doi = {10.7717/peerj-cs.3964}
}
EUVIMED