AI carbon footprint in US Central - Iowa: emissions per token and per query

Factor version soma-ai-ef-2026.11-dataset-v3.1.0. Factors last updated . Location-based grid averages, not renewable contracts. Estimates from published benchmarks.

Page updated

Part of the SOMA AI Emission Factors cluster.

How much CO₂ does an AI query emit when served from US Central - Iowa?

A 1210-token query to a class B model (gpt-4o) served from US Central - Iowa emits about 0.052 g CO₂e, and a million tokens emit 0.0381 kg CO₂e.

Rank 14 of 16 for class B, lowest carbon first (full table). 5.3x the level of EU North - Sweden, Asia Pacific - Singapore is 1.1x higher.

Which grid supplies the electricity in US Central - Iowa?

US Central - Iowa maps to MRO West, a US eGRID subregion, at 472.1 g CO₂e per kWh for 2023. That is 2.0 times the EU-27 average (236 g) and 1.0 times the global average (483.2 g). The source is eGRID2023 MROW 926.6 lb/MWh CO2e total output x 1.1232 (Ember lifecycle basis).

Cloud regions on this grid, 2 in all (1 Azure, 1 Google Cloud).

Grid MROW.
ProviderRegion idRegion nameLocationMapping confidenceMapping note
AzurecentralusCentral USIowahighProvider states state only; polygon test at West Des Moines (41.60,-93.70) = MROW. Same subregion as GCP us-central1 (Council Bluffs). CONFIRMED 2026-10-14 by Data Center Map (datacentermap.com, operator pages for Microsoft and Amazon AWS, read 14 October 2026): Microsoft Ginger East campus, West Des Moines (29 listings); polygon test at West Des Moines (41.58,-93.71) = MROW; agrees with the assumed site, confidence raised from medium to high.
Google Cloudus-central1us-central1Council Bluffs, IowahighSubregion settled in paper (anchor).

The note for Google Cloud us-central1 reads "Subregion settled in paper (anchor).".

Which grids have an intensity close to MRO West?

Closest grids to 472.1 g CO₂e per kWh, with the ratio to the MRO West figure.
Difference gGridCodeg CO₂e per kWhRatio
2.6Viet NamVNM474.71.006
-5.4RFC WestRFCW466.70.989
7.9Cabo VerdeCPV4801.017
-11.0ASCC Alaska GridAKGD461.10.977
11.7United Arab EmiratesARE483.81.025
-11.9SERC Tennessee ValleySRTV460.20.975
12.8MacaoMAC484.91.027
16.9MexicoMEX4891.036
19.1CambodiaKHM491.21.040
-19.4GhanaGHA452.70.959

Emission factors for US Central - Iowa by model class

Per million tokens unless stated, per query = 1210 tokens.
Model classElectricity kgHardware kgConstruction kgTraining kgTotal kg CO₂eg per queryWater LmL per query
Class A (small)0.010280.000270.000140.000600.0110.0160.0690.096
Class B (mid-size)0.033650.001010.000460.002930.0380.0520.2270.31
Class C (frontier)0.125320.003880.001690.012840.1440.20.8461.2
kg CO₂e per million tokens, development shown apart.
Model classCentralClass range lowClass range highHigh caseB200Development
Class A0.011290.0069720.016360.018110.009890.01401
Class B0.038050.024530.046190.060420.026650.04589
Class C0.14370.056540.21510.22710.14780.1709

How much does siting matter for a model served from US Central - Iowa?

Electricity is 88.4% of the class B total here (class A 91%, class C 87.2%), with hardware at 2.7%, construction 1.2% and training 7.7%. Electricity carries most of the 0.0381 kg, so the grid is the biggest lever.

The example company (2,323 million class B tokens a year) emits 88 kg CO₂e from US Central - Iowa. Moving to US South - Southeast would save 6.8 kg (7.7%) or US South Central - Texas would save 16 kg (18%).

Kilograms CO₂e a year saved by moving the example traffic from US Central - Iowa to each region, by model class.
RegionGrid g per kWhClass AClass BClass C
US South - Southeast4312.086.825.3
US South Central - Texas375.34.91659.7

How much water does AI usage consume in US Central - Iowa?

A class B query of 1210 tokens from US Central - Iowa consumes about 0.31 mL of water and a million tokens 0.227 litres, or 0.126 litres without the water attributed to hydropower.

Water inputs.
InputValue
On-site WUE, L per kWh0.123
WUE denominatortotal site electricity
WUE basis as publishedwithdrawal
Withdrawal to consumption0.6667
EWIF, L per kWh3.104
EWIF without hydro, L per kWh1.680
AWARE factor3.41
Confidencederived-site
SourceMicrosoft; Des Moines (IA) (site); FY25
Water per million tokens in litres.
Model classOn siteGrid electricityTotalTotal without hydroAWARE-weighted on site
Class A0.001780.06760.06930.03840.00607
Class B0.005830.2210.2270.1260.0199
Class C0.02170.8240.8460.4680.0741

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, 12 November 2026, CC BY 4.0, concept DOI 10.5281/zenodo.20443585, which resolves to the latest version (this version 10.5281/zenodo.24160409, 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, accepted for publication in Sustainable Production and Consumption.

Partners can fetch any row from the AI factors API, and the whole table is a CSV download.

Cite as "0.0381 kg CO₂e per million tokens, class B, US Central - Iowa, SAEF, 12 November 2026" and link somaai.earth/ai-footprint/factors/us-central.

Keep reading