AI Energy and CO₂ Calculator (GPU Training or Inference)
Enter GPU power and hours to estimate energy, cost and CO₂.
Your numbers
Your results
AI Energy and CO₂ Calculator (GPU Training or Inference) by TheCalcTools
How the AI Energy and CO₂ Calculator (GPU Training or Inference) works
Energy = power × GPUs × hours, multiplied by the data centre’s power usage effectiveness (PUE), which adds cooling and power distribution. Emissions multiply energy by the grid’s carbon intensity.
GPU power varies with load; use average draw, not the maximum rating, if you know it. Server CPUs, memory and networking add more.
The formula
Worked example
Using the default values shown in the calculator:
| Average power per GPU (W) | 700 |
|---|---|
| Number of GPUs | 64 |
| Hours | 720 |
| Data centre PUE | 1.2 |
| Electricity price per kWh | $0.12 |
| Grid carbon intensity (g CO₂ per kWh) | 400 |
- 700 W × 64 GPUs × 720 h × 1.2 = 38.7 MWh.
- Cost about $4,645; emissions about 15.5 t CO₂ at 400 g/kWh.
Result:
| Energy (MWh) | 38.71 |
|---|---|
| Electricity cost | $4,645 |
| CO₂ emissions (tonnes) | 15.48 |
| Equivalent average EU household-years (3,500 kWh) | 11.1 |
Frequently asked questions
What is a typical grid carbon intensity?
It varies hugely: under 50 g/kWh in hydro- or nuclear-heavy grids, roughly 200 to 500 in many mixed grids and over 700 in coal-heavy ones. Use a current figure for your region.
Sources and further reading
Related calculators
More in AI & Computing · All calculators
This calculator gives estimates for information only. Check important results independently before you rely on them. See our disclaimer.