DOE Well-to-Wheel Methodology Applied to AI Compute
Well-to-Inference Calculator
Like "well-to-wheel" for vehicles tracks emissions from oil extraction through combustion,Well-to-Inference tracks AI's carbon footprint from power generation through model inference—with cryptographic receipts at every stage.
In Genesis terms: This is the receipt primitive that lets you tie every scientific inference back to where the electrons came from, when they were used, and under which policy they were allowed to run.
Automotive (DOE Standard)
WellRefineryPumpWheel
AI Compute (Our Method)
GenerationTransmissionData CenterInference
Topology:Linear (A → B → C → D)
|Configure Scenario
Permian Pilot — Genesis proving ground in West Texas (TTU/ERCOT region)
Carbon Accumulation Pipeline
STAGE 1
Generation
385
gCO₂/kWh
STAGE 2
Transmission
404
gCO₂/kWh
STAGE 3
Data Center
566
gCO₂/kWh eff.
STAGE 4
Inference
2.04
gCO₂/inf
Total Energy
3.60
kWh
Total Emissions
2037
gCO₂
Per Inference
2.04
gCO₂/inference