EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

The Carbon Additionality Frontier for Agentic AI in Sustainable Business

Kwan Hong Tan

Journal of Artificial Intelligence and Sustainable Business · 16.11.2026 · 1 (1) · 42-66

DOI: 10.68607/jaisb.v1i1.17 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Artificial Intelligence and Sustainable Business, online erschienen am 16.11.2026. Verfasst von Kwan Hong Tan. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Agentic artificial intelligence can reduce the emissions intensity of business activities while increasing their total environmental burden through expanded activity, repeated computation and verification. This conceptual paper develops the carbon additionality frontier, a decision framework that jointly constrains deployment scale and the evidence needed to substantiate carbon savings. A normalised model separates expected operational emissions, fixed deployment emissions, verification emissions, reducible uncertainty and an uncertainty floor. Analytical propositions establish a maximum admissible activity multiplier, a bounded interval of verification effort and conditions under which no amount of additional checking can support a carbon-saving claim. For a specified uncertainty-contraction function, the model yields a closed-form verification optimum and extends to general convex verification costs and uncertainty sets. Reproducible numerical illustrations show how an apparently large improvement in emissions intensity can support only a narrow conservative saving after expansion and assurance are included. A worked example yields a 31.85% activity-expansion ceiling and a 2.80% conservative saving at 28% expansion; these are conditional model outputs, not empirical estimates. The paper translates the frontier into deployment gates, supplier evidence requirements and a prospective empirical design. Its contribution is to treat sustainable AI deployment as a joint problem of physical performance, evidence quality and enforceable scale. Carbon efficiency becomes an input to a bounded decision, rather than sufficient evidence of sustainable growth.

Zitieren

APA

Tan, K. H. (2026). The Carbon Additionality Frontier for Agentic AI in Sustainable Business. Journal of Artificial Intelligence and Sustainable Business, 1(1), 42-66. https://doi.org/10.68607/jaisb.v1i1.17

Vancouver

Tan KH. The Carbon Additionality Frontier for Agentic AI in Sustainable Business. Journal of Artificial Intelligence and Sustainable Business. 2026;1(1):42-66. doi:10.68607/jaisb.v1i1.17

BibTeX
@article{Tan2026,
  author = {Tan, Kwan Hong},
  title = {The Carbon Additionality Frontier for Agentic AI in Sustainable Business},
  journal = {Journal of Artificial Intelligence and Sustainable Business},
  year = {2026},
  volume = {1},
  number = {1},
  pages = {42--66},
  doi = {10.68607/jaisb.v1i1.17}
}
Kostenlos starten →