The frame
A Flow With an Inventory
Water is produced, held in a tank, and drawn down by several claimants at once. Managing that one tank against many demands is the discipline: allocation.
Water on a site is a flow like energy and compute - mined or recovered, stored, and consumed on a running balance. What sets it apart is that the store is a single inventory with many claimants pulling on it, and the plant filling it cannot be run flat out. So the governing problem is not how to make water but how to divide it, hour by hour, among uses that each believe they come first.
Six commodities
Six Commodities in One Molecule
One substance does the work of six supply chains, and the fact that it is one substance is exactly what makes allocation hard.
| Role | What it supplies |
|---|---|
| Propellant & oxidizer | Hydrogen and oxygen for engines, split by electrolysis - the export and the return ticket |
| Breathing gas | Crew oxygen, from the same electrolysis step |
| Crew supply | Drinking, food preparation, hygiene |
| Shielding mass | Hydrogen-rich mass over a storm shelter - the best stopping power per tonne available |
| Energy store | The regenerative fuel-cell charge that carries a site across the dark |
| Heat-transfer fluid | The working fluid in the thermal loop and the radiators |
Six claimants, one tank. Because they draw on the same molecule, a tonne spent as propellant is a tonne not available as shielding or fuel-cell charge - the uses are not additive, they are rivals.
Allocation
A Site Runs Out of Unallocated Water
Closure is an allocation problem, not a recycling problem. The number that matters is not how much water exists but how much is unspoken for.
A site with a working mine and a full tank can still fail if every tonne is already promised. The binding quantity is unallocated water - what remains after propellant, oxygen, crew supply, shielding, the fuel-cell charge, and the thermal loop have each taken their claim, against a plant that cannot run at peak continuously. A site does not run out of water; it runs out of unallocated water. That reframes the whole problem from recycling efficiency, which is a plumbing question, to allocation policy, which is a command question.
The store is the product. The regenerative fuel-cell charge is about 4.3 tonnes of water - seven days of a 230 t/yr plant’s output - held back as the energy buffer, so the deepest claim on the inventory is the one that keeps the lights on. The allocation policy that arbitrates all six claims has no owner yet; it belongs as a design freeze before a site is committed, not after.
Grade
Grade, Not Technology, Sets the Cost
The energy to win a tonne of water is dominated by ore grade, and most of it is wasted heating rock that is then thrown away.
Extracting water costs between about 1 and 4 kWh per kilogram, and the spread is set by grade - the water fraction of the deposit - far more than by the extraction method. A third to a half of that energy is sensible heat, spent warming regolith that is then discarded once its water is driven off. That single fact names the highest-leverage improvement in the chain: beneficiation, concentrating the ore before the thermal step so less dead rock is heated. It has never been costed. The grade of the permanently shadowed lunar ice is itself a named falsifier, because it swings the water rung’s energy by nearly a factor of three and decides whether water and oxygen production can share one power bus. Projected
The working fluid
Whether the Export Exists Is Levied on the Customer
Water is the only heat-transfer fluid a lunar site can make, which quietly makes the Moon's top export depend on someone else's spacecraft design.
A radiator rejects heat through a working fluid running in heat pipes, and at the 320 to 350 K band a lunar industrial site operates in, that fluid is water. Ammonia, the usual spacecraft choice, cannot be manufactured on the Moon. A 110 m² radiator is about 340 m of heat pipe at a 16 cm fin length, all of it carrying water. So whether the Moon can sell cooling and propellant at all is a requirement levied on the customer’s vehicle - its thermal system has to accept a water-based fluid - years before either the plant or the customer exists. The export is real only if that design choice is made upstream. Projected
Closure
Water Either Closes or It Does Not
Unlike energy, which always closes somehow, water closure is close to binary and set by a single geological fact.
Energy closes everywhere by some route - solar, then fission. Water does not. A site either sits on an accessible deposit or it does not, and that one fact decides whether water closes or stays an import on a resupply clock. Past resupply range, a place without accessible water is not a destination for people; it is somewhere robots visit. That binary is what puts water at the center of the Presence Gate - the test that sorts each destination into crewed, contested, or uncrewed turns, more than on anything else, on whether the water is there. Building
Contested
Deliberately Not Claimed
The load-bearing uncertainties, tagged with the confidence they earn.
The six uses and the allocation framing are solid. Each role is demonstrated on operating spacecraft and stations; that one molecule serves all six is chemistry, not conjecture.
Lunar ice grade is the swing variable. The permanently shadowed deposits are known to exist but not characterized for grade, and the grade sets the extraction energy within a factor of three - which decides what else fits on the power bus.
Beneficiation is the unstudied lever. Concentrating ore before the thermal step is where the largest energy saving sits, and no one has costed it, so the water-energy figures carry that gap.
Record
Provenance and Data Sources
- Confidence
- Observed for the six uses and the allocation framing - Building for the energy figures and closure logic - Projected for lunar ice grade and the export working-fluid dependency
- Status date
- Q4 2026
- Discipline
- Allocation - one inventory, six claimants, a plant that cannot run flat out; a site runs out of unallocated water, not water
- Key figures
- Six roles in one molecule; extraction 1–4 kWh/kg, grade-dominated, a third to a half wasted as sensible heat; fuel-cell charge ~4.3 t = seven days held; radiator fluid at 320–350 K
- Related
- Energy for the fuel-cell buffer; Materials for bulk matter as stock; the Presence Gate that water decides
- Weakest link
- Lunar ice grade and the uncosted beneficiation step - both move the water-and-oxygen power budget