AI Carbon Footprint Calculator

Estimate the carbon emissions of an AI/ML workload (training or inference) using average power draw, runtime, and grid intensity. This is a directional estimate for planning.

Workload inputs

Runtime (hours)Average power draw per device (W)Number of devices (GPUs/TPUs/servers)Average utilization (%)Data center PUEGrid intensity (kg CO₂e / kWh)

Formula: Energy (kWh) = watts * utilization * hours * devices / 1000. Facility energy applies PUE. Emissions = facility kWh * grid intensity.

Results

Estimated total emissions

0.0000 tCO₂e

(0.0 kg CO₂e)

IT energy0.00 kWh

Facility energy (PUE applied)0.00 kWh

Devices1

Utilization70%

Grid intensity0.250 kg/kWh

Tip

For more accurate results, use measured wall power (or cloud provider energy reports) and a region-specific grid factor.

This tool estimates operational emissions only; it does not include embodied emissions from hardware manufacturing.