AI carbon footprint in EU Central - Germany: 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 Central - Germany?
A 1210-token query to a class B model (gpt-4o) served from EU Central - Germany emits about 0.041 g CO₂e, and a million tokens emit 0.0303 kg CO₂e.
Rank 11 of 16 for class B, lowest carbon first (full table). 4.2x the level of EU North - Sweden, Asia Pacific - Singapore is 1.3x higher.
Which grid supplies the electricity in EU Central - Germany?
EU Central - Germany maps to Germany, a national grid, at 363.6 g CO₂e per kWh for 2023. That is 1.5 times the EU-27 average (236 g) and 0.8 times the global average (483.2 g). The source is Ember yearly electricity data, CO2 intensity 2023, 363.59 g/kWh (lifecycle, IPCC AR5).
Cloud regions on this grid, 6 in all (2 AWS, 2 Azure, 2 Google Cloud).
| Provider | Region id | Region name | Location | Mapping confidence | Mapping note |
|---|---|---|---|---|---|
| AWS | eu-central-1 | Europe (Frankfurt) | Frankfurt | high | |
| AWS | eusc-de-east-1 | AWS European Sovereign Cloud (Brandenburg) | Brandenburg | medium | GA 2026-01-14 per web search results (AWS blog page itself not fetched). Separate partition (aws-eusc), not on the official global regions list; public commercial but sovereign-operated. Delete this row if sovereign partitions are out of scope. |
| Azure | germanynorth | Germany North | Berlin | 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. |
| Azure | germanywestcentral | Germany West Central | Frankfurt | high | |
| Google Cloud | europe-west10 | europe-west10 | Berlin, Germany | high | |
| Google Cloud | europe-west3 | europe-west3 | Frankfurt, Germany | high |
Which grids have an intensity close to Germany?
| Difference g | Grid | Code | g CO₂e per kWh | Ratio |
|---|---|---|---|---|
| -3.7 | Belarus | BLR | 359.9 | 0.990 |
| -3.8 | WECC Southwest | AZNM | 359.8 | 0.990 |
| -9.0 | Rwanda | RWA | 354.6 | 0.975 |
| 9.2 | Tanzania, the United Republic of | TZA | 372.8 | 1.025 |
| 11.7 | ERCOT All | ERCT | 375.3 | 1.032 |
| -12.7 | Argentina | ARG | 350.9 | 0.965 |
| 14.3 | SERC Mississippi Valley | SRMV | 377.9 | 1.039 |
| -14.6 | Pakistan | PAK | 349 | 0.960 |
| -16.7 | Faroe Islands (the) | FRO | 346.9 | 0.954 |
| -17.4 | Korea (the Democratic People's Republic of) | PRK | 346.2 | 0.952 |
Emission factors for EU Central - Germany 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) | 0.00792 | 0.00027 | 0.00014 | 0.00060 | 0.009 | 0.012 | 0.044 | 0.061 |
| Class B (mid-size) | 0.02592 | 0.00101 | 0.00046 | 0.00293 | 0.030 | 0.041 | 0.144 | 0.2 |
| Class C (frontier) | 0.09652 | 0.00388 | 0.00169 | 0.01284 | 0.115 | 0.16 | 0.537 | 0.74 |
| Model class | Central | Class range low | Class range high | High case | B200 | Development |
|---|---|---|---|---|---|---|
| Class A | 0.008926 | 0.005561 | 0.01288 | 0.01421 | 0.007845 | 0.01079 |
| Class B | 0.03032 | 0.01977 | 0.03663 | 0.04764 | 0.0215 | 0.03535 |
| Class C | 0.1149 | 0.04698 | 0.1705 | 0.1795 | 0.1181 | 0.1316 |
How much does siting matter for a model served from EU Central - Germany?
Electricity is 85.5% of the class B total here (class A 88.7%, class C 84%), with hardware at 3.3%, construction 1.5% and training 9.7%. Electricity carries most of the 0.0303 kg, so the grid is the biggest lever.
The example company (2,323 million class B tokens a year) emits 70 kg CO₂e from EU Central - Germany. Moving to US West - Oregon would save 6.6 kg (9.4%) or US East - N. Virginia would save 9.9 kg (14%).
| Region | Grid g per kWh | Class A | Class B | Class C |
|---|---|---|---|---|
| US West - Oregon | 323.7 | 2.02 | 6.61 | 24.6 |
| US East - N. Virginia | 303.8 | 3.02 | 9.9 | 36.9 |
How much water does AI usage consume in EU Central - Germany?
A class B query of 1210 tokens from EU Central - Germany consumes about 0.2 mL of water and a million tokens 0.144 litres, or 0.127 litres without the water attributed to hydropower.
| Input | Value |
|---|---|
| On-site WUE, L per kWh | 0.17 |
| WUE denominator | IT load |
| WUE basis as published | withdrawal |
| Withdrawal to consumption | 0.6667 |
| EWIF, L per kWh | 1.930 |
| EWIF without hydro, L per kWh | 1.681 |
| AWARE factor | 2.09 |
| Confidence | published-site |
| Source | AWS; AWS Region Europe (Frankfurt) eu-central-1; 2025 |
| Model class | On site | Grid electricity | Total | Total without hydro | AWARE-weighted on site |
|---|---|---|---|---|---|
| Class A | 0.00206 | 0.042 | 0.0441 | 0.0386 | 0.0043 |
| Class B | 0.00673 | 0.138 | 0.144 | 0.127 | 0.0141 |
| Class C | 0.0251 | 0.512 | 0.537 | 0.471 | 0.0524 |
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 "0.0303 kg CO₂e per million tokens, class B, EU Central - Germany, SAEF, 19 October 2026" and link somaai.earth/ai-footprint/factors/eu-germany.