Midjourney Environmental Impact

Energy Consumption

Midjourney uses approximately 7.5 Wh per generation cycle (producing 4 images in 30-60 seconds). At ~1300W total system power for H100 inference, a 45-second generation consumes approximately 16 Wh, with conservative estimates at 5-10 Wh per cycle. In Q2 2025, Midjourney migrated most of its inference fleet from NVIDIA A100/H100 clusters to Google Cloud TPU v6e pods, achieving a 65% cost reduction — which likely also reduced energy consumption per image compared to its previous GPU-based infrastructure.

Power Source & Carbon

Midjourney runs on Google Cloud Platform, using both Google TPUs and NVIDIA GPU-based VMs. Google Cloud claims 100% renewable energy matching and uses custom-designed TPUs that are more power-efficient than general-purpose GPUs for AI workloads. The specific regions are not publicly disclosed but likely include US Central, US East, and Europe West.

Water Usage

At approximately 18 mL per image, Midjourney's water footprint is comparable to DALL-E 3. Google's data centers use both evaporative cooling and air-side cooling depending on the climate. Google has committed to replenishing 120% of the water it consumes across its operations by 2030.

How much energy does Midjourney use per query?

Each Midjourney query consumes approximately 7.5 Wh of energy. This is 25x more than a traditional Google search (~0.3 Wh).

What is Midjourney's carbon footprint?

Based on the carbon intensity of Google Cloud (US/EU TPU regions), each query produces approximately 2.3 g of CO2. The grid in this region has a carbon intensity of 300 g CO2/kWh with 64% renewable energy.

How much water does Midjourney use?

Each query consumes approximately 14 mL of water, primarily used for cooling the data centers that process the request.

How does Midjourney compare to a Google search?

A Midjourney query uses 25x more than a Google search in terms of energy. A Google search uses approximately 0.3 Wh, while Midjourney uses 7.5 Wh.

Architecture

Diffusion model (proprietary)

Release date

2023-12-21

Open source

No