AI Footprint

AI's Environmental Footprint: Beyond the Myths

You've probably heard conflicting claims about AI's environmental impact: some alarming, some dismissive.

We're here to show you what 2025 first-party data actually reveals, help you estimate your own usage, and support organisations that need audit-ready reporting.

Prompt

Estimate the environmental impact of your AI text prompts.

Calculate Prompt Impact

Asset

Estimate resources for AI assets and analyse the impact of various AI output types.

Estimate Asset & Output

Action

Get an AI-powered estimation for custom or complex AI tasks.

Estimate AI Action

Did You Know?

Latest 2025 disclosures: ~0.3 Wh per AI text query. That's 10× more efficient than 2023 estimates. See the methodology.

Scale reference: ~50 typical AI prompts use the same electricity as one smartphone charge (15 Wh).

Your AI footprint depends on three things: model size, task type, and data centre location. The calculator estimates all three.

Figures from first-party 2025-26 disclosures and the sourced references in the "Understanding AI's Environmental Footprint" section below.

Understanding AI's Environmental Footprint
Learn how generative AI affects our planet, based on the best available first-party disclosures.

Tips for Sustainable AI Usage
Now that you understand what drives AI's environmental impact, here's how to reduce it. These strategies work across all AI tools and use cases.
  • Optimise Your Prompts

    Write clear, concise prompts to get the information you need with fewer iterations, reducing computational load.

  • Choose Smaller Models When Possible

    For simpler tasks, smaller, more efficient models can be just as effective whilst using fewer resources.

  • Be Mindful of Task Intensity

    Text generation typically uses far less energy than generating images or videos. Use intensive tasks judiciously.

  • Consider Location & Provider

    Support AI services hosted in regions with cleaner energy grids and by companies committed to renewable energy.

  • Minimise Redundant Usage

    Avoid unnecessary AI interactions. Think if the task truly requires AI or if a simpler tool would suffice.