Calculators
Open models for the numbers that decide an AI build. Defaults are drawn from the register — change them to your case.
GPU TCO & cost-per-GPU-hour
$2.18 / GPU-hr
Annual cost / GPU$16,265
· Depreciation$12,500
· Energy$765
· Opex$3,000
Rental ladder (2026): spot ~$1.03 · neocloud ~$2.29–3.50 · AWS ~$6.88 · Azure ~$12.29. Below the ~70% utilization breakeven a debt-financed cluster bleeds cash → Ch 1.8.
Inference cost per million tokens
$1.9 / M tokens
Self-hosted worked example. Market self-serve fell ~$10 → ~$2.50 / M tokens in a year (LLMflation ~10×/yr) — underwrite inference with a price-decline curve, never flat → Ch 1.8.
Facility energy & water
Facility draw120 MW
Annual energy1,051,200 MWh
Annual energy cost$84,096,000
Annual water350.4 M L
PUE bands: legacy air 1.4–1.6 · direct-to-chip liquid 1.05–1.15 → Ch 15.1.
Save, compare, and export scenarios are coming with accounts. The project-finance / levered-IRR model lands in a later pass.