OCEANS™ AI Footprint
Estimate the climate footprint of your AI workloads.
Translate AI activity into estimated electricity use and greenhouse-gas emissions using transparent assumptions you can review and adjust.
What this estimates
This calculator supports three kinds of AI workload:
- · Token-based inference — running a trained model to generate text.
- · Image generation — producing images from a generative model.
- · Training or fine-tuning — measured in GPU-hours.
How the estimate works
Energy (kWh) = a per-unit energy factor × your quantity × PUE (data-center overhead). Emissions (kg CO₂e) = energy × the carbon intensity of the electricity grid you select.
Uncertainty statement
These are illustrative default factors, not a measured, audited, or certified result. Published estimates for AI energy use vary widely by model size, hardware, and implementation — the low/central/high range reflects that spread, not a precise measurement of any specific system. Replace the defaults with your own numbers if you have them.
Methodology
Version 0.1.0
Estimated result
Low
0.288 kWh
0.138 kg CO₂e
Central
0.720 kWh
0.346 kg CO₂e
High
1.15 kWh
0.553 kg CO₂e
Central estimate in tonnes: 0.000346 tCO₂e
- Grid factor
- 0.48 kg CO₂e/kWh
- PUE applied
- 1.2
- Methodology
- v0.1.0
Reading the number is only the first step