AI carbon footprint in EU West - France: emissions per token and per query
Factor version soma-ai-ef-2026.10-dataset-v3.0.0. Last updated . Location-based grid averages, not renewable contracts. Estimates from published benchmarks.
Part of the SOMA AI Emission Factors cluster.
How much CO₂ does an AI query emit when served from EU West - France?
A 1210-token query to a class B model (gpt-4o) served from EU West - France emits about 0.011 g CO₂e, and a million tokens emit 0.0082 kg CO₂e.
Rank 2 of 16 for class B, lowest carbon first (full table). 1.1x the level of EU North - Sweden, Asia Pacific - Singapore is 4.9x higher.
Which grid supplies the electricity in EU West - France?
EU West - France maps to France, a national grid, at 53.3 g CO₂e per kWh for 2023. That is 0.2 times the EU-27 average (236 g) and 0.1 times the global average (483.2 g). The source is Ember yearly electricity data, CO2 intensity 2023, 53.26 g/kWh (lifecycle, IPCC AR5).
Cloud regions on this grid, 4 in all (1 AWS, 2 Azure, 1 Google Cloud).
| Provider | Region id | Region name | Location | Mapping confidence | Mapping note |
|---|---|---|---|---|---|
| AWS | eu-west-3 | Europe (Paris) | Paris | high | |
| Azure | francecentral | France Central | Paris | high | |
| Azure | francesouth | France South | Marseille | high | Access restricted by Microsoft to specific customer scenarios (e.g. disaster recovery); listed on the official page but not open to general new subscriptions. |
| Google Cloud | europe-west9 | europe-west9 | Paris, France | high |
Which grids have an intensity close to France?
| Difference g | Grid | Code | g CO₂e per kWh | Ratio |
|---|---|---|---|---|
| 1.3 | Malawi | MWI | 54.6 | 1.024 |
| 5.6 | Uganda | UGA | 58.9 | 1.105 |
| -5.7 | Namibia | NAM | 47.6 | 0.893 |
| -5.7 | Sierra Leone | SLE | 47.6 | 0.893 |
| -14.9 | Sweden | SWE | 38.4 | 0.720 |
| -18.5 | Switzerland | CHE | 34.8 | 0.653 |
| 21.1 | Tajikistan | TJK | 74.4 | 1.396 |
| -22.8 | Norway | NOR | 30.5 | 0.572 |
| -25.5 | Iceland | ISL | 27.8 | 0.522 |
| -26.4 | Congo (the Democratic Republic of the) | COD | 26.9 | 0.505 |
Emission factors for EU West - France by model class
| Model class | Electricity kg | Hardware kg | Construction kg | Training kg | Total kg CO₂e | g per query | Water L | mL per query |
|---|---|---|---|---|---|---|---|---|
| Class A (small) | 0.00116 | 0.00027 | 0.00014 | 0.00060 | 0.002 | 0.0029 | 0.081 | 0.11 |
| Class B (mid-size) | 0.00380 | 0.00101 | 0.00046 | 0.00293 | 0.008 | 0.011 | 0.264 | 0.36 |
| Class C (frontier) | 0.01414 | 0.00388 | 0.00169 | 0.01284 | 0.033 | 0.043 | 0.982 | 1.4 |
| Model class | Central | Class range low | Class range high | High case | B200 | Development |
|---|---|---|---|---|---|---|
| Class A | 0.002171 | 0.001525 | 0.002933 | 0.003045 | 0.001995 | 0.001581 |
| Class B | 0.008198 | 0.006127 | 0.009684 | 0.01109 | 0.006772 | 0.005178 |
| Class C | 0.03256 | 0.01965 | 0.04285 | 0.04338 | 0.03308 | 0.01928 |
How much does siting matter for a model served from EU West - France?
Electricity is 46.3% of the class B total here (class A 53.4%, class C 43.4%), with hardware at 12%, construction 5.6% and training 36%. A cleaner grid shrinks the electricity term, while the other 53.7% stays put.
The example company (2,323 million class B tokens a year) emits 19 kg CO₂e from EU West - France. Moving to EU North - Sweden would save 2.5 kg (13%).
| Region | Grid g per kWh | Class A | Class B | Class C |
|---|---|---|---|---|
| EU North - Sweden | 38.4 | 0.751 | 2.46 | 9.15 |
How much water does AI usage consume in EU West - France?
A class B query of 1210 tokens from EU West - France consumes about 0.36 mL of water and a million tokens 0.264 litres, or 0.149 litres without the water attributed to hydropower.
| Input | Value |
|---|---|
| On-site WUE, L per kWh | 0.05 |
| WUE denominator | IT load |
| WUE basis as published | withdrawal |
| Withdrawal to consumption | 0.6667 |
| EWIF, L per kWh | 3.671 |
| EWIF without hydro, L per kWh | 2.061 |
| AWARE factor | 4.99 |
| Confidence | continental |
| Source | AWS; AWS Europe aggregate; 2025 |
| Model class | On site | Grid electricity | Total | Total without hydro | AWARE-weighted on site |
|---|---|---|---|---|---|
| Class A | 0.000605 | 0.0799 | 0.0805 | 0.0455 | 0.00302 |
| Class B | 0.00198 | 0.262 | 0.264 | 0.149 | 0.00988 |
| Class C | 0.00737 | 0.975 | 0.982 | 0.554 | 0.0368 |
Where do these numbers come from and how do I cite them?
Figures per query are one plain exchange, before the Tier 2b reasoning uplift. Electricity term. GPU energy per token from the ML.ENERGY Leaderboard v3, a facility overhead of 1.20, and grid carbon intensity from EPA eGRID 2023 and Ember 2023. Embodied hardware, data-centre construction and amortised training terms are documented on the methodology page. Water combines on-site cooling water (Li et al. 2025) with the water embedded in grid electricity (Reig et al. 2020). The modelled range of the models inside a class, around the class value, is −41% to +48% for class A, −39% to +24% for class B and −67% to +55% for class C.
Dataset: AI Inference Emission and Resource Factors for Corporate GHG Inventories (SOMA AI Emission Factors (SAEF)), Llopis-Garcia, G., SOMA AI, 19 October 2026, CC BY 4.0, concept DOI 10.5281/zenodo.20443585, which resolves to the latest version (this version 10.5281/zenodo.23371447, earlier version 2 values at 10.5281/zenodo.22767475). Method: Accounting for Purchased AI Inference in Corporate GHG Inventories: A Token-Based Method, arXiv:2606.10660, under peer review.
Partners can fetch any row from the AI factors API, and the whole table is a CSV download.
Cite as "0.0082 kg CO₂e per million tokens, class B, EU West - France, SAEF, 19 October 2026" and link somaai.earth/ai-footprint/factors/eu-france.