Skip to content

Energy data licenses and attributions

Venturi’s energy, carbon, and water estimates are built on open methodologies and public datasets. This page credits those upstream sources and states exactly what Venturi reuses, reimplements, and does not redistribute.

The guiding rule is simple: Venturi reimplements published methodology from primary sources and preserves every upstream license on any data it redistributes. Where a source’s data is not licensed for redistribution, Venturi does not redistribute it; it reimplements the method from the primary description instead.

AI Energy Score

The AI Energy Score is a standardized, multi-organization model-energy benchmark maintained by Hugging Face with Salesforce, Cohere, and Carnegie Mellon University. Venturi bundles a versioned copy of its published coefficients and ratings as the primary energy tier (T1) and records the benchmark’s run-config provenance with every coefficient, as described in the energy methodology. The benchmark source is MIT; the current leaderboard Space declares Apache-2.0. The AI Energy Score name and label are used as reference material, and standalone ratings link back to the source leaderboard.

CodeCarbon: MIT License

CodeCarbon is an open-source emissions-tracking library released under the MIT License. Venturi uses CodeCarbon as the optional measured-energy backend (tier T0) for self-hosted deployments (ingesting its emissions.csv, BoAmps JSON, and Prometheus metrics) and adopts its documented carbon-resolution method (power = Δenergy / Δhours, the five-step grid-intensity chain, and the world-average fallback). CodeCarbon’s own bundled datasets carry their own upstream terms (see Our World in Data and cloud-provider intensities below), which are preserved on any redistribution.

Our World in Data: grid carbon intensity

Per-country grid carbon-intensity and generation-mix figures originate from Our World in Data, as vendored through CodeCarbon’s grid-intensity datasets. Where Venturi redistributes these per-country intensities in its catalog, the Our World in Data attribution and the source dataset’s license are preserved. The current OWID Energy dataset declares CC BY 4.0. Cloud-provider per-region intensities retain their own source links and terms.

How-Hungry-is-AI: methodology reimplemented, data not redistributed

The two-term water model, the infra-aware energy estimate (tier T2), the eco-efficiency DEA, and the real-world equivalence divisors are reimplemented from the methodology published in How Hungry is AI? Benchmarking Energy, Water, and Carbon Footprint of LLM Inference, which is released under CC BY 4.0. Under CC BY 4.0 the paper’s formulas, constants, and method are reusable with attribution, which this page provides.

What Venturi does not redistribute

Venturi does not redistribute the How-Hungry-is-AI project’s CSV datasets or its dashboard. Those artifacts are research/informational-use material, not licensed for commercial redistribution. Venturi reimplements the methodology from the CC-BY-4.0 paper and recomputes its own numbers from primary sources (the AI Energy Score, CodeCarbon, provider sustainability disclosures, and public benchmark results); it does not ship the How-Hungry-is-AI data files.

Electricity Maps: customer-supplied licensed feed

Electricity Maps is an optional carbon-intensity source. A customer supplies a token under its own subscription and selected use type. Venturi does not bundle, resell, or publicly redistribute Electricity Maps data. Use remains subject to the current Electricity Maps terms, and every resolved value records Electricity Maps as its source.

Eco-efficiency and provider sustainability inputs

The fixed AI-Index weights and DEA cross-efficiency calculation are Venturi methodology. The DEA framing follows the CC BY 4.0 How Hungry paper; benchmark scores retain their publishers’ terms and source URLs. Provider PUE, WUE, and carbon-intensity inputs are reimplemented from published provider sustainability disclosures and are versioned with provider, region, source URL, and vintage.

Summary

Source Used for License / basis Redistributed?
AI Energy Score T1 energy coefficients + ratings MIT source; Apache-2.0 leaderboard Space Coefficients vendored, named, linked back
CodeCarbon T0 measured-energy ingestion + carbon method MIT Method adopted; bundled data per its upstream terms
Our World in Data Per-country grid intensity Current Energy dataset CC BY 4.0 Yes, with attribution and source license preserved
How-Hungry-is-AI Water / infra-aware / eco-efficiency methodology + equivalences Paper under CC BY 4.0 Methodology only: CSVs and dashboard not redistributed
Electricity Maps Optional live grid intensity Customer subscription and selected use type No public redistribution

Where to go next