A MetroVolt fleet runs on helium-3, and the series treats supply as an engineering system, not a hand-wave: measured consumption, staged sources, and an in-plant contribution computed against its own fuel cycle.
A unit consumes helium-3 at reactor scale (fleet planning uses ≈90 kg/yr class figures at full buildout in the series' economics). Sources are staged: terrestrial tritium-decay stockpiles (³He accrues at 0.42 liters per gram of tritium per year — the series' corrected, deposited figure), D-D-derived breeding inside the staged fuel cycle itself, and market/strategic reserves. The self-supply fraction is computed honestly: f_self ≈ 0.34 rising toward 0.48 across the staging, with the in-situ supply trajectory plotted from the deposited fuel-cycle integrator.
The series is blunt where it must be: prompt in-plant breeding at ratio ≥1 does not close (§3.5 computes it negative) — self-sufficiency is partial by physics, which is exactly why the supply strategy is staged and external sourcing is a permanent line item.
Fuel security is a solvable procurement problem when consumption is honest and sources are plural. The plan's shape — stockpile decay + partial self-breeding + reserves — is resilient to any single source disappointing, and every input to that plan is deposited where a fuel buyer can audit it.
| Decay sourcing | 0.42 L ³He / g tritium / yr (deposited) |
| Self-supply fraction | f_self ≈ 0.34 → 0.48 (staged) |
| Prompt breeding ≥ 1? | no — computed negative, stated (§3.5) |
| Fleet-scale demand class | ≈90 kg/yr at buildout |
| Strategy | stockpiles + partial self-breeding + reserves |