Overview
Methodology for Cloud AI Token Inference Energy, Carbon, and Water Measurement
Greenpixie's AI token methodology provides per-token energy and embodied emissions estimates for cloud AI inference, and estimates the water consumption associated with that inference, covering large language models and embedding models hosted on AWS, Azure, and GCP. It is a companion to the Cloud methodology.
Start with the Introduction, or jump to any section from the sidebar.
1. Introduction
Purpose, audience, and context of the methodology.
2. Scope and Applicability
What is in scope, out of scope, and the evaluation boundary.
3. Definitions
Key terms used throughout the methodology.
4. Methodology Overview
How the methodology works at a high level.
5. Inputs and Evidence Sources
The data categories and evidence sources the methodology draws on.
6. Eligibility, Selection, and Inclusion Logic
What is included and how it is selected and mapped.
7. Assessment and Calculation
The calculation components and their outputs.
8. Monitoring, Updates, and Review Timing
How the methodology is kept current.
9. Quality Assurance and Oversight
Standards alignment, validation, and review controls.
10. Governance and Methodology Changes
Working group, review cadence, and version control.
11. Outputs and Interpretation
Primary outputs, intended use, and interpretation guidance.
12. Related Documents
Standards, research, and sources referenced by the methodology.
13. Notice on Intellectual Property and Permitted Use
IP notice and permitted use of this document.