Skip to content

#11 of 30 · Cloud + AI lab

Amazon (AWS / Nova)

The lowest disclosed water intensity, inside the largest and fastest-growing footprint.

Seattle, Washington·amazon.com ·Verified Oct 16, 2026
BGood confidence
Index score
Good confidence

In plain words

How scores work

Amazon scores 57 out of 100 — middle of the pack. Some disclosure or clean-energy commitments, offset by rising emissions or blind spots. Confidence in this score is good confidence (B).

What helps

  • Carbon · Matched 100% of electricity with renewable energy for a third year on an annual, global basis; net-zero target 2040.
  • Water · Data-centre WUE of 0.15 L/kWh in 2024, a 17% improvement year on year and the lowest disclosed figure in this index.
  • Water · 53% of the way to returning more water than AWS operations use by 2030.

What hurts

  • Efficiency · No per-prompt or per-model energy figures for Nova or Bedrock; efficiency claims are hardware-level (Trainium2, PUE 1.14) only.
  • Carbon · Absolute emissions rose 16% in 2025 to 80.85 MtCO₂e — the largest single-year increase since Amazon began tracking — with Scope 2 up 34%, mainly data centres.
  • Carbon · Absolute energy consumption for AWS is still not disclosed; the report is limited to efficiency metrics.

What nobody knows

  • Per-prompt energy disclosed
  • Per-prompt water disclosed
  • Training emissions in model cards

AWS hosts Anthropic's Claude on Trainium (Project Rainier), serves Amazon's own Nova models and much of the industry's inference. Its 2024 WUE of 0.15 L/kWh is the lowest disclosed in the index and it has matched 100% of electricity with renewables three years running — but absolute emissions jumped 16% in 2025 to 80.9 MtCO₂e, the biggest rise since it began reporting, Scope 2 rose 34% on data centres, and AWS still does not disclose how much electricity it uses.

Products
Amazon Bedrock · Amazon Nova · Trainium / Inferentia · Alexa+
Hosting
Own AWS regions worldwide; Project Rainier (≈500,000 Trainium2 chips) across US sites including Indiana.

Key figures

Per-prompt and fleet metrics. Hover a title for what it means; open “How we got this number” for the working.

How to read the labels:Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Measured independentlyMeasured independently. Measured or estimated by researchers or analysts with no stake in the result.ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Not disclosedNot disclosed. The company publishes nothing on this and no defensible estimate exists.— hover any label.
Energy per promptElectricity used to answer one typical request. Under 0.3 Wh is good. A 10 W LED bulb uses 0.3 Wh in about two minutes.
0.25
Wh / typical prompt
≈ 20.8 smartphone charges per 1,000 prompts
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.AmazonLuccioni, Jernite, Strubell (arXiv 2311.16863)Epoch AI

ecogpt.co estimate; no company figure.

How we got this number

Amazon's Nova models are mid-size and served on its own Trainium and Inferentia chips; that puts them between the measured sub-10B median (0.04 Wh) and frontier models (≈0.3 Wh). Uncertainty roughly ±3×.

Water per promptWater evaporated cooling the servers for one request. Scope matters: on-site cooling only, or also the water used to generate the electricity (which adds ≈1.8 mL per Wh).
0.04
mL / typical prompt
≈ 0.07 500 mL bottles per 1,000 prompts
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.AmazonLuccioni, Jernite, Strubell (arXiv 2311.16863)Epoch AIAmazon

On-site cooling only, so comparable with Google and OpenAI's figures.

How we got this number

0.25 Wh per prompt × 0.15 L/kWh (AWS's disclosed fleet WUE). Counting the water evaporated to generate the electricity adds ≈0.45 mL.

Carbon per promptGreenhouse gas from one request on the grid where it actually runs. Below 0.1 g is clean. A kilometre in a petrol car is about 170 g.
0.098
gCO₂e / typical prompt
≈ 0.58 km driven (petrol car) per 1,000 prompts
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.AmazonLuccioni, Jernite, Strubell (arXiv 2311.16863)Epoch AI

Location-based: what the grid actually emitted. Renewable certificates would bring the market-based figure near zero.

How we got this number

0.25 Wh × 390 gCO₂/kWh (US grid, location-based).

Annual emissionsThe company's whole yearly footprint, in million tonnes of CO₂e. Clouds and device makers dwarf pure labs because they also count shops, logistics or products. Compare the trend, not the size.
80.9
MtCO₂e (2025)
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Data Center Dynamics

Whole company, including retail and logistics.

Emissions trendDirection of travel against last year or the company's own baseline. Negative is good. Positive means the AI build-out is outrunning clean energy.
+16.0
% YoY (2025)
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Data Center Dynamics

Scope 2 +34%, Scope 3 +20%.

Water per kWh (WUE)Litres of water consumed per kilowatt-hour of server electricity, across the fleet. 0.15 is excellent, 0.3 is good, 1.8 is the US data-centre average.
0.15
L/kWh (2024)
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Amazon
Clean electricityShare of electricity matched with renewable or carbon-free sources. Annual matching counts certificates bought anywhere; hourly matching on the same grid is far stricter.
100
% renewable matching (market-based)
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Data Center Dynamics

Annual and global, not hourly.

Hardware & materials

The chips, power, copper and future e-waste behind the fleet — the part sustainability reports leave out

Factors we use
AI chips in service
2.5 million
H100-equivalents (Epoch AI, Q4 2025)
≈ 34,722 racks of 72 GPUs
Measured independentlyMeasured independently. Measured or estimated by researchers or analysts with no stake in the result.Network WorldAmazon

GPUs, TPUs or custom accelerators the company runs. Analyst figures are in H100-equivalents, company figures in physical chips.

How we got this number

Roughly half AMD GPUs, half AWS Trainium; Project Rainier alone is ≈500,000 Trainium2 chips.

Power draw
3,250
MW
≈ 2.4 million US homes' worth of power
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldAmazonSemiAnalysis

Megawatts the AI fleet pulls. One megawatt runs roughly 750 US homes.

How we got this number

2,500,000 accelerators × 1.3 kW each, all-in. Fleet power implied by the chip count.

Electricity a year
22.8
TWh / year
≈ 2.2 million US homes for a year
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldAmazonSemiAnalysis

Terawatt-hours per year. One TWh is a year of electricity for about 95,000 US homes.

How we got this number

3250 MW × 8,760 h × 80% utilisation. Compute fleet only.

Carbon baked into the hardware
1,285
ktCO₂e embodied in chips and servers
≈ 279,348 petrol cars driven for a year
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldAmazonarXiv 2607.05408NVIDIA developer blog

CO₂e emitted making the chips and servers before they answer a single prompt.

How we got this number

2,500,000 accelerators × 514 kg CO₂e each (chip plus share of the host server). Emitted in manufacturing, before the first prompt is served.

Future e-waste
42,500
tonnes of servers to retire within ~5 years
≈ 236 empty Boeing 747s by weight
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldAmazonNVIDIAThe Register

Tonnes of server hardware that will be retired in three to five years.

How we got this number

2,500,000 accelerators × 17 kg of server hardware each. Tomorrow's e-waste unless reused; generative AI could produce 2.5 Mt of it a year by 2030.

Copper in the build
87,750
tonnes of copper in the build
≈ copper for 1.1 million electric cars
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldAmazonSemiAnalysisBHPS&P Global Market Intelligence

Tonnes of copper in cabling, busbars and transformers.

How we got this number

3,250 MW × 27 t of copper per MW (cabling, busbars, transformers).

Fossil generators on site
1,250
backup diesel generators (implied)
≈ 3,250 MW of backup capacity
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldAmazonSemiAnalysisLatitude MediaLatitude Media / Floodlight

Diesel backup units or gas turbines at the data centres, permitted or observed.

How we got this number

3,250 MW ÷ 2.6 MW per generator, the average size permitted at US data centres. Permit counts are not published fleet-wide; this is the number a fleet this size normally carries.

Hardware reused or recycled
—
AWS runs re:Cycle reverse-logistics hubs but publishes no fleet reuse share.
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Amazon

Share of retired servers and parts given a second life.

Mineral depletion
—
Only Mistral has published a mineral-depletion figure.
Not disclosedNot disclosed. The company publishes nothing on this and no defensible estimate exists.

Kilograms of antimony-equivalent, the life-cycle unit for scarce minerals consumed.

Score breakdown

Each pillar is 0–100; weights are fixed across the index

Energy efficiency — How much electricity one answer costs.weight 20% · confidence D
50
  • No per-prompt or per-model energy figures for Nova or Bedrock; efficiency claims are hardware-level (Trainium2, PUE 1.14) only. AmazonAmazon
Carbon intensity — How clean that electricity is, and whether total emissions are rising.weight 25% · confidence B
48
  • Absolute emissions rose 16% in 2025 to 80.85 MtCO₂e — the largest single-year increase since Amazon began tracking — with Scope 2 up 34%, mainly data centres. Data Center Dynamics
  • Matched 100% of electricity with renewable energy for a third year on an annual, global basis; net-zero target 2040. Data Center DynamicsAmazon
  • Absolute energy consumption for AWS is still not disclosed; the report is limited to efficiency metrics. Data Center Dynamics
Water — How much water the data centres consume, and whether it is replaced.weight 15% · confidence B
85
  • Data-centre WUE of 0.15 L/kWh in 2024, a 17% improvement year on year and the lowest disclosed figure in this index. Amazon
  • 53% of the way to returning more water than AWS operations use by 2030. Amazon
Transparency — How much the company actually publishes.weight 20% · confidence B
50
  • Annual report with PUE 1.14 and WUE; regions published; no AI-specific disclosures and no AWS electricity total. AmazonData Center Dynamics
Commitments — Whether it has dated, measurable climate and water targets.weight 10% · confidence A
65
  • Net-zero carbon by 2040 (Climate Pledge) and water positive by 2030. Amazon
Conduct — How it behaves: permits, lawsuits, gas turbines, community water.weight 10% · confidence C
60
  • No AI-specific enforcement actions on record; rapid Rainier build-out in Indiana relies on outside-air cooling per AWS. Amazon

Models

What this company ships, and what each prompt costs

ModelWeightsEnergy / promptWaterTraining CO₂e
Amazon Nova Pro
Mid-size (undisclosed)
Closed0.25 Wh Estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.0.037 mL Estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.3,000 t Estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.

Where the compute lives

Named sites from company disclosures and press reporting

OperationalConstructionPlanned
  • Northern Virginia (us-east-1)Virginia, USA
  • Project RainierSt. Joseph County, Indiana, USA · ≈500,000 Trainium2 chips for Anthropic.
  • Umatilla / BoardmanOregon, USA
  • Dublin (eu-west-1)Ireland
  • SingaporeSingapore

Timeline

Incidents and developments involving this company

HighEmissions·Jul 1, 2026

Amazon's emissions jump 16% to 80.9 Mt — its largest rise ever — on record data-centre additions

Amazon's 2025 sustainability report shows absolute emissions of 80.85 MtCO₂e, up 16% in a year, with Scope 2 up 34% on data-centre electricity. AWS still does not disclose how much electricity it uses; the report is limited to PUE (1.14) and WUE.

Amazon
NotableEmissions·Jul 16, 2025

Amazon's emissions rise 6% to 68.25 Mt while AWS posts the lowest water intensity disclosed

Absolute emissions grew from 64.38 to 68.25 MtCO₂e in 2024 as carbon intensity per dollar fell 4%. AWS reported PUE 1.15 and WUE 0.15 L/kWh, a 17% improvement.

Amazon