AI carbon footprint in EU South - Spain: 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 South - Spain?
A 1210-token query to a class B model (gpt-4o) served from EU South - Spain emits about 0.022 g CO₂e, and a million tokens emit 0.0165 kg CO₂e.
Rank 5 of 16 for class B, lowest carbon first (full table). 2.3x the level of EU North - Sweden, Asia Pacific - Singapore is 2.4x higher.
Which grid supplies the electricity in EU South - Spain?
EU South - Spain maps to Spain, a national grid, at 169.7 g CO₂e per kWh for 2023. That is 0.7 times the EU-27 average (236 g) and 0.4 times the global average (483.2 g). The source is Ember yearly electricity data, CO2 intensity 2023, 169.68 g/kWh (lifecycle, IPCC AR5).
Cloud regions on this grid, 3 in all (1 AWS, 1 Azure, 1 Google Cloud).
| Provider | Region id | Region name | Location | Mapping confidence | Mapping note |
|---|---|---|---|---|---|
| AWS | eu-south-2 | Europe (Spain) | Spain | high | Provider states country only. |
| Azure | spaincentral | Spain Central | Madrid | high | |
| Google Cloud | europe-southwest1 | europe-southwest1 | Madrid, Spain | high |
The note for AWS eu-south-2 reads "Provider states country only.".
Which grids have an intensity close to Spain?
| Difference g | Grid | Code | g CO₂e per kWh | Ratio |
|---|---|---|---|---|
| -2.0 | Georgia | GEO | 167.7 | 0.988 |
| 4.7 | Canada | CAN | 174.4 | 1.028 |
| 10.3 | Burundi | BDI | 180 | 1.061 |
| -11.3 | Portugal | PRT | 158.4 | 0.933 |
| 13.8 | Angola | AGO | 183.5 | 1.081 |
| -14.1 | Belize | BLZ | 155.6 | 0.917 |
| 15.3 | Guinea | GIN | 185 | 1.090 |
| -17.4 | Denmark | DNK | 152.3 | 0.897 |
| -23.6 | Sudan (the) | SDN | 146.1 | 0.861 |
| 25.2 | Croatia | HRV | 194.9 | 1.148 |
Emission factors for EU South - Spain 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) | 0.00369 | 0.00027 | 0.00014 | 0.00060 | 0.005 | 0.0064 | not available | not available |
| Class B (mid-size) | 0.01210 | 0.00101 | 0.00046 | 0.00293 | 0.016 | 0.022 | not available | not available |
| Class C (frontier) | 0.04504 | 0.00388 | 0.00169 | 0.01284 | 0.063 | 0.086 | not available | not available |
| Model class | Central | Class range low | Class range high | High case | B200 | Development |
|---|---|---|---|---|---|---|
| Class A | 0.004705 | 0.003039 | 0.006664 | 0.007232 | 0.004189 | 0.005037 |
| Class B | 0.0165 | 0.01124 | 0.01965 | 0.02481 | 0.0123 | 0.01649 |
| Class C | 0.06347 | 0.02991 | 0.09073 | 0.09443 | 0.06498 | 0.06142 |
How much does siting matter for a model served from EU South - Spain?
Electricity is 73.3% of the class B total here (class A 78.5%, class C 71%), with hardware at 6.1%, construction 2.8% and training 18%. Electricity carries most of the 0.0165 kg, so the grid is the biggest lever.
The example company (2,323 million class B tokens a year) emits 38 kg CO₂e from EU South - Spain. Moving to EU West - Belgium would save 5.8 kg (15%) or EU North - Finland would save 15 kg (38%).
| Region | Grid g per kWh | Class A | Class B | Class C |
|---|---|---|---|---|
| EU West - Belgium | 134.7 | 1.77 | 5.8 | 21.6 |
| EU North - Finland | 81.3 | 4.47 | 14.6 | 54.5 |
How much water does AI usage consume in EU South - Spain?
Water: not available for EU South - Spain. The dataset publishes no off-site water factor for this region, so SOMA prints no water figure rather than a partial one.
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 "0.0165 kg CO₂e per million tokens, class B, EU South - Spain, SAEF, 19 October 2026" and link somaai.earth/ai-footprint/factors/eu-spain.