Skip to content

#25 of 30 · Image-generation product

Midjourney

Every image costs five to ten chat prompts of electricity; nothing is disclosed.

San Francisco, California·midjourney.com ·Verified Oct 16, 2026
DLittle data
Index score
Little data

In plain words

How scores work

Midjourney scores 36 out of 100 — laggard. Little or no disclosure, fossil-backed capacity, or active disputes with regulators and communities. Confidence in this score is little data (D).

What helps

  • Efficiency · Trains on Google TPUs and reportedly moved inference to TPU v6e in 2025, cutting cost by about two thirds, which implies an efficiency gain.

What hurts

  • Efficiency · Image generation measured at 1.6 Wh (AI Energy Score, SDXL) to 3.6 Wh per image at 1024 px — far above a text prompt.
  • Water · Inherits Google's fleet WUE of 1.15 L/kWh — high by hyperscaler standards — on an energy-hungry workload.
  • Transparency · No environmental disclosure of any kind.

What nobody knows

  • Per-prompt energy disclosed
  • Per-prompt water disclosed
  • Training emissions in model cards
  • Annual environmental report
  • Location-based Scope 2 reported
  • +4 more

Midjourney generates images for more than 20 million members on Google Cloud, training on TPUs and reportedly moving inference from NVIDIA GPUs to TPU v6e in 2025. Independent measurements put a 1024-pixel diffusion image at 1.6–3.6 Wh — five to ten times a text prompt — but Midjourney has never published energy, water or emissions figures, and has no climate targets.

Products
Midjourney V7 · Niji · Midjourney Video
Hosting
Google Cloud (TPU v4/v6e for training and inference; NVIDIA GPU VMs).

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.
2.90
Wh / image
≈ 242 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.Hugging Face, Salesforce, Cohere, Carnegie Mellonheise onlineLuccioni, Jernite, Strubell (arXiv 2311.16863)

ecogpt.co estimate; Midjourney publishes nothing.

How we got this number

Hugging Face and Carnegie Mellon measured 1.64 Wh per Stable Diffusion XL image at default settings and up to 3.58 Wh at 1024×1024 pixels. Midjourney renders at 1024 px by default and upscales, so 2.9 Wh is the mid-to-upper end. Uncertainty ±2×.

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).
3.34
mL / typical prompt
≈ 6.67 500 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.Hugging Face, Salesforce, Cohere, Carnegie Mellonheise onlineLuccioni, Jernite, Strubell (arXiv 2311.16863)Google (arXiv 2508.15734)

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

How we got this number

2.9 Wh per prompt × 1.15 L/kWh (Google Cloud's disclosed fleet WUE). Counting the water evaporated to generate the electricity adds ≈5.22 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.
1.131
gCO₂e / typical prompt
≈ 6.65 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.Hugging Face, Salesforce, Cohere, Carnegie Mellonheise onlineLuccioni, Jernite, Strubell (arXiv 2311.16863)

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

How we got this number

2.9 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.
0.01
MtCO₂e / year (AI compute only)
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.Google Cloud

Location-based estimate for the AI fleet only; not comparable with a full corporate inventory.

How we got this number

4 MW × 8,760 h × 80% utilisation × 390 gCO₂/kWh (US grid). Compute only — nothing else the company does is counted.

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.
+30.0
% (member growth)
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.Google Cloud
How we got this number

No inventory exists; membership has grown roughly a third a year since the 11 million members Google cited in 2023.

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.
1.15
L/kWh (inherited from Google Cloud)
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google (arXiv 2508.15734)
How we got this number

Midjourney runs on Google Cloud, whose fleet WUE is 1.15 L/kWh.

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.
66.0
% hourly carbon-free (inherited from Google Cloud)
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google
How we got this number

Google's fleet average hourly carbon-free share in 2024; Midjourney has no procurement of its own.

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
3,000
GPU-equivalents (order of magnitude)
≈ 42 racks of 72 GPUs
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.Google Cloud

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

How we got this number

Reported inference spend of about $2.1M a month on NVIDIA GPUs before a 2025 move to TPU v6e works out to roughly 1,500 H100-equivalents running flat out; doubled for training and peaks. Uncertainty ±3×.

Power draw
4.00
MW
≈ 3,000 US homes' worth of power
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google CloudSemiAnalysis

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

How we got this number

3,000 accelerators × 1.3 kW each, all-in. Fleet power implied by the chip count.

Electricity a year
0.03
TWh / year
≈ 2,857 US homes for a year
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google CloudSemiAnalysis

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

How we got this number

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

Carbon baked into the hardware
1.5
ktCO₂e embodied in chips and servers
≈ 326 petrol cars driven for a year
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google CloudarXiv 2607.05408NVIDIA developer blog

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

How we got this number

3,000 accelerators × 514 kg CO₂e each (chip plus share of the host server). Emitted in manufacturing, before the first prompt is served.

Future e-waste
51.0
tonnes of servers to retire within ~5 years
≈ 0.3 empty Boeing 747s by weight
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google CloudNVIDIAThe Register

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

How we got this number

3,000 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
108
tonnes of copper in the build
≈ copper for 1,301 electric cars
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google CloudSemiAnalysisBHPS&P Global Market Intelligence

Tonnes of copper in cabling, busbars and transformers.

How we got this number

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

Fossil generators on site
2.00
backup diesel generators (implied)
≈ 5.2 MW of backup capacity
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Google CloudSemiAnalysisLatitude MediaLatitude Media / Floodlight

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

How we got this number

4 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
—
No hardware reuse or e-waste figure published.
Not disclosedNot disclosed. The company publishes nothing on this and no defensible estimate exists.

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 C
35
Carbon intensity — How clean that electricity is, and whether total emissions are rising.weight 25% · confidence C
55
  • Runs entirely on Google Cloud: 100% annual renewable matching, roughly 66% hourly carbon-free. GoogleGoogle Cloud
Water — How much water the data centres consume, and whether it is replaced.weight 15% · confidence C
45
  • Inherits Google's fleet WUE of 1.15 L/kWh — high by hyperscaler standards — on an energy-hungry workload. Google (arXiv 2508.15734)
Transparency — How much the company actually publishes.weight 20% · confidence D
5
Commitments — Whether it has dated, measurable climate and water targets.weight 10% · confidence C
10
Conduct — How it behaves: permits, lawsuits, gas turbines, community water.weight 10% · confidence C
60

Models

What this company ships, and what each prompt costs

ModelWeightsEnergy / promptWaterTraining CO₂e
Midjourney V7 (per image)
Diffusion (undisclosed)
Closed2.90 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.3.335 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.500 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
  • San Francisco HQCalifornia, USA
  • Google Cloud (us-central1)Council Bluffs, Iowa, USA · TPU and GPU capacity.