What this indicator measures
This indicator estimates electricity used by AI workloads, including compute activity and supporting infrastructure where applicable.
Calculation method
The model converts estimated AI workload activity into electricity consumption using hardware and infrastructure assumptions.
Simplified formula
Electricity use ≈ AI compute workload × hardware power factor × infrastructure overheadData sources
Sources may include AI infrastructure research, GPU specifications, data center efficiency assumptions, operator disclosures and public energy statistics.
Published calculation inputs
The values below are the explicit inputs currently used by this indicator. They are assumptions or public proxies, not official totals.
- Baseline
- 560 TWh/a
- Model assumption
Direct references
Last reviewed: September 27, 2026
Known limitations
The exact electricity use of AI systems is rarely disclosed. The estimate depends on assumptions about hardware utilization, model size, inference volume, training activity and data center efficiency.
Refresh and interpolation
The server periodically refreshes the baseline. The live display interpolates electricity consumption between refreshes.