AI carbon footprint in EU West - Belgium: 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 - Belgium?
A 1210-token query to a class B model (gpt-4o) served from EU West - Belgium emits about 0.019 g CO₂e, and a million tokens emit 0.014 kg CO₂e.
Rank 4 of 16 for class B, lowest carbon first (full table). 2.0x the level of EU North - Sweden, Asia Pacific - Singapore is 2.9x higher.
Which grid supplies the electricity in EU West - Belgium?
EU West - Belgium maps to Belgium, a national grid, at 134.7 g CO₂e per kWh for 2023. That is 0.6 times the EU-27 average (236 g) and 0.3 times the global average (483.2 g). The source is Ember yearly electricity data, CO2 intensity 2023, 134.67 g/kWh (lifecycle, IPCC AR5).
Cloud regions on this grid, 2 in all (1 Azure, 1 Google Cloud).
| Provider | Region id | Region name | Location | Mapping confidence | Mapping note |
|---|---|---|---|---|---|
| Azure | belgiumcentral | Belgium Central | Brussels | high | |
| Google Cloud | europe-west1 | europe-west1 | St. Ghislain, Belgium | high |
Which grids have an intensity close to Belgium?
| Difference g | Grid | Code | g CO₂e per kWh | Ratio |
|---|---|---|---|---|
| 0.7 | El Salvador | SLV | 135.4 | 1.005 |
| -3.5 | Greenland | GRL | 131.2 | 0.974 |
| 5.0 | Kyrgyzstan | KGZ | 139.7 | 1.037 |
| -5.9 | Mozambique | MOZ | 128.8 | 0.956 |
| -11.0 | Afghanistan | AFG | 123.7 | 0.918 |
| -11.0 | NPCC Upstate NY | NYUP | 123.7 | 0.918 |
| 11.4 | Sudan (the) | SDN | 146.1 | 1.085 |
| -13.5 | Eswatini | SWZ | 121.2 | 0.900 |
| -14.1 | Luxembourg | LUX | 120.6 | 0.895 |
| -15.6 | Latvia | LVA | 119.1 | 0.884 |
Emission factors for EU West - Belgium 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) | 0.00293 | 0.00027 | 0.00014 | 0.00060 | 0.004 | 0.0054 | 0.057 | 0.08 |
| Class B (mid-size) | 0.00960 | 0.00101 | 0.00046 | 0.00293 | 0.014 | 0.019 | 0.188 | 0.26 |
| Class C (frontier) | 0.03575 | 0.00388 | 0.00169 | 0.01284 | 0.054 | 0.073 | 0.701 | 0.97 |
| Model class | Central | Class range low | Class range high | High case | B200 | Development |
|---|---|---|---|---|---|---|
| Class A | 0.003943 | 0.002583 | 0.005542 | 0.005973 | 0.00353 | 0.003998 |
| Class B | 0.014 | 0.009705 | 0.01665 | 0.02068 | 0.01064 | 0.01309 |
| Class C | 0.05417 | 0.02682 | 0.07634 | 0.07908 | 0.05538 | 0.04875 |
How much does siting matter for a model served from EU West - Belgium?
Electricity is 68.6% of the class B total here (class A 74.4%, class C 66%), with hardware at 7.2%, construction 3.3% and training 21%. Electricity carries most of the 0.014 kg, so the grid is the biggest lever.
The example company (2,323 million class B tokens a year) emits 33 kg CO₂e from EU West - Belgium. Moving to EU North - Finland would save 8.8 kg (27%) or EU West - France would save 13 kg (41%).
| Region | Grid g per kWh | Class A | Class B | Class C |
|---|---|---|---|---|
| EU North - Finland | 81.3 | 2.7 | 8.83 | 32.9 |
| EU West - France | 53.3 | 4.12 | 13.5 | 50.2 |
How much water does AI usage consume in EU West - Belgium?
A class B query of 1210 tokens from EU West - Belgium consumes about 0.26 mL of water and a million tokens 0.188 litres, or 0.158 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 | 2.612 |
| EWIF without hydro, L per kWh | 2.195 |
| AWARE factor | 3.52 |
| 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 | 0.0569 | 0.0575 | 0.0484 | 0.00213 |
| Class B | 0.00198 | 0.186 | 0.188 | 0.158 | 0.00697 |
| Class C | 0.00737 | 0.693 | 0.701 | 0.59 | 0.026 |
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 "0.014 kg CO₂e per million tokens, class B, EU West - Belgium, SAEF, 19 October 2026" and link somaai.earth/ai-footprint/factors/eu-belgium.