The conversion factor
Every Kilowatt Costs 0.37 Tons of Energy Chain
55 t of generation, storage, distribution and heat rejection delivers 150 kWe at the lunar minimum viable outpost.
The factor converts any load into landed mass. It comes from the bill of materials in The Primitives and prices every rung of the closure ladder; it is why the ladder is ordered by energy.
| Load | Average power | Energy chain | Rated |
|---|---|---|---|
| The Machine Shop MVP, average | 6.7 kW | 2.5 t | Modeled |
| Oxygen, 100 t a year | 80 kW | 30 t | Modeled |
| Water, 100 t a year at 5 wt% ice | 57 kW | 21 t | Modeled |
| Lunar minimum viable outpost, whole bus | 150 kWe | 55 t | Modeled |
| Metals reduction, 100 t a year | 228-457 kW | 84-167 t | Modeled |
Specific energy
Seven Orders of Magnitude From Moving Dirt to Making a Chip
Sorting transformations by energy per kilogram puts the rungs in order without further assumption; each waits for the power plant to grow about tenfold.
| Rung | Transformation | kWh per kg of product | Basis | Rated |
|---|---|---|---|---|
| Bulk mass | Excavate, haul, emplace regolith | 0.0003-0.001 | 0.3-1.0 kWh per ton moved | Modeled |
| Salvage | Cut and re-form a landed hull | About 0.04 | Plasma cut on 4 mm plate | Modeled |
| Structure | Melt silicate; cast, sinter or draw fiber | 0.53 ideal, 0.84 realized | 1,500 K rise plus fusion | Modeled |
| Water | Heat icy regolith | 1.1-4.1 ideal, 3-14 realized | Ice grade 10% down to 1% | Modeled |
| Oxygen | Electrolyze water | 4.97 ideal, about 7 realized | 286 kJ per mole of water | Observed |
| Metals | Reduce oxide to metal | 20-40 | Molten regolith electrolysis | Modeled |
| Devices | Semiconductor process chain | 1,000-100,000 | Fab energy per kg of device | Modeled |
Salvage sits off the curve: cutting a landed Starship hull liberates finished 304L at 0.04 kWh/kg, material that carries 10-15 kWh/kg of terrestrial embodied energy.
Unlock cost
What Must Land Before a Rung Produces Anything
Dedicated hardware plus the energy chain its average draw implies. The column does not sum: the rungs share one bus and take turns.
| Rung | Hardware | Average power | Energy chain | Produces | Rated |
|---|---|---|---|---|---|
| Salvage | About 0.2 t | About 5 kW | 1.8 t | 66 t of 304L stainless per hull | Modeled |
| Bulk mass | 23.5 t | About 50 kW | 18.4 t | Thousands of m3 a year | Modeled |
| Water | About 12 t | 57-114 kW | 21-42 t | 100 t a year | Modeled |
| Oxygen | About 10 t | About 80 kW | 30 t | 100 t a year | Modeled |
| Spares by machining | 3.1 t | 6.7 kW | 2.5 t | 69% of failures | Modeled |
| Structure, forming | About 1.1 t | 10-25 kW installed | About 4 t | Tube, gaskets, bearing blanks | Modeled |
| Structure, fiber line | 2.5-5 t | 10-25 kW installed | 4-9 t | Mineral fiber from local silicate | Modeled |
| Metals reduction | No credible figure | 228-457 kW | 84-167 t | 100 t of alloy a year | Modeled |
The energy chains total 76-102 t against an outpost energy layer of 55 t, so the rungs are dispatched in campaigns, not run together - see one bus on The Triple Substrate.
Closure ratio
Below the Spares Rung, Closure Pays in Mass; Above It, in Schedule
Closure ratio = annual product mass ÷ (plant mass + 0.37 t/kWe × average power), per year. Its reciprocal is the payback time.
| Rung | Closure ratio | Payback | Mass test | Rated |
|---|---|---|---|---|
| Salvage | About 22 per year | About 3 weeks | Passes | Modeled |
| Bulk mass | Undefined | - | Not applicable; displaces nothing | Modeled |
| Oxygen | About 2.5 per year | About 5 months | Passes | Modeled |
| Water | 1.9-3.0 per year | 4-6 months | Passes; spread is ice grade | Modeled |
| Spares by machining | Marginal at the Moon, negative beyond | 7.5 years at the Moon | Fails | Modeled |
| Structure, forming | About 5 per build | One campaign | Passes | Modeled |
| Structure, cast base | 1.3-2.6 per expansion | Second expansion | Passes, and compounds | Modeled |
| Metals reduction | Under 1 | Never at outpost demand | Fails as closure; passes only as export | Modeled |
Salvage, oxygen, water and structure return their own landed mass within months to one campaign. Spares by machining fail the mass test everywhere past the Moon, and metals reduction fails it at any outpost demand, so their case rests on availability, design-loop time and self-expansion, as The Machine Shop MVP shows. Metals reduction is an export rung, justified only when the customer is somebody else.
Sunlight
Array Area Grows 90-Fold From Earth to Saturn
Sunlight falls with the square of distance; array area for 100 kWe at 25% end-to-end efficiency.
| Destination | Distance | Sunlight | Array per 100 kWe | Rated |
|---|---|---|---|---|
| Mercury | 0.39 AU | 9,090 W/m2 | 44 m2 | Modeled |
| Venus orbit | 0.72 AU | 2,600 W/m2 | 154 m2 | Modeled |
| Earth orbit, Moon | 1.0 AU | 1,361 W/m2 | 294 m2 | Modeled |
| Mars | 1.52 AU | 586 W/m2 | 683 m2 | Modeled |
| Ceres | 2.77 AU | 177 W/m2 | 2,260 m2 | Modeled |
| Jupiter system | 5.20 AU | 50 W/m2 | 7,950 m2 | Modeled |
| Saturn system | 9.54 AU | 15 W/m2 | 26,700 m2 | Modeled |
Area is at full sun, before night, dust and angle. A lunar equatorial array is dark half the time; south-pole ridges are lit 80-90% of the year; Mars loses output for weeks in dust storms. Past the Belt, every serious plan uses a reactor.
Heat rejection
A Kilowatt Needs About 1.2 m2 of Radiator in Shadow and Three Times That in Lunar Daylight
Rejection depends on radiator temperature to the fourth power and on what the panel sees.
| Environment | Effective sink | At 300 K | At 350 K | Rated |
|---|---|---|---|---|
| Deep space; lunar polar shadow | 3-40 K | 1.21 m2 | 0.65 m2 | Modeled |
| Callisto surface | About 120 K | 1.24 m2 | 0.66 m2 | Modeled |
| Mars surface | About 200 K | 1.51 m2 | 0.73 m2 | Modeled |
| Lunar day, vertical panel over hot ground | About 270 K | 3.52 m2 | 1.01 m2 | Modeled |
Running hotter cuts area steeply: 350 K needs about half the area of 300 K. Electronics limit how hot a compute radiator can run, which is why cold sinks - lunar polar shadow, Callisto - matter for compute.
Storage
Past About Seven Hours of Darkness, Water Beats Batteries
Li-ion at 175 Wh/kg against a regenerative fuel cell with 25 kg/kW fixed mass and about 1.9 kg per kWh of water and tankage.
| Darkness | Li-ion at 175 Wh/kg | Regenerative fuel cell | Lighter | Rated |
|---|---|---|---|---|
| 1 h | 0.1 t | 0.7 t | Battery | Modeled |
| 6.6 h | 0.9 t | 0.9 t | Crossover | Modeled |
| 10 h | 1.4 t | 1.1 t | Fuel cell | Modeled |
| 100 h, best lunar polar ridge | 14 t | 5.3 t | Fuel cell | Modeled |
| 354 h, equatorial lunar night | 51 t | 17 t | Fuel cell | Modeled |
At a site making its own water, the fuel-cell reactant is local and only stacks and tanks ship. The same tank is also an oxygen and propellant reserve, which is why reserves are physically separated.
Water by grade
Ice Grade Moves Water Power Nearly Threefold Between 5% and 1%
Ideal heat is the regolith warmed 200 K plus 0.79 kWh/kg to sublimate the ice; realized is 2.5-3.5 times ideal.
| Ice grade | Regolith per kg water | Ideal heat | Realized | Power per 100 t a year | Rated |
|---|---|---|---|---|---|
| 10 wt% | 9 kg | 1.1 kWh/kg | 2.7-3.8 kWh/kg | 31-44 kW | Modeled |
| 5 wt% (LCROSS at Cabeus: 5.6) | 19 kg | 1.4 kWh/kg | 3.6-5.0 kWh/kg | 41-57 kW | Observed |
| 2 wt% | 49 kg | 2.4 kWh/kg | 6.1-8.5 kWh/kg | 69-97 kW | Modeled |
| 1 wt% | 99 kg | 4.1 kWh/kg | 10-14 kWh/kg | 117-163 kW | Modeled |
One impact, LCROSS at Cabeus in 2009, is the only direct measurement of lunar polar ice grade. CNSA's Chang'e-7 is built to make the second.
Exchange rates
Energy Extracts Water, Water Stores Energy, Compute Turns Power to Heat
The conversions behind The Triple Substrate.
| Conversion | Rate | Rated |
|---|---|---|
| Electrolysis, water to hydrogen and oxygen | 4.4 kWh/kg water ideal; 6.3 at 70% | Observed |
| Fuel cell, back to electricity | About 2 kWh per kg water cycled | Modeled |
| Oxygen yield | 0.89 kg O2 per kg water | Observed |
| Compute power to heat | 1 W in, 1 W out | Observed |
| Crew breathing oxygen | 0.84 kg O2 per person per day | Observed |
| Crew CO2 output | About 1 kg per person per day | Observed |
Light time
Light Time Sets Who Decides
Round-trip signal time against the speed of the events that matter.
| Destination | Round trip | Control mode | Rated |
|---|---|---|---|
| Earth orbit | Under 0.3 s | Direct teleoperation | Observed |
| Moon | 2.6 s | Teleoperation with lag | Observed |
| Mars | 6-44 min | Daily plans; local reaction | Observed |
| Ceres | 30-62 min | Plans per day or week | Observed |
| Callisto | 66-106 min | Rules and approvals only | Observed |
| Pluto | 8-14 h | Full autonomy | Observed |
A Mars landing takes about seven minutes against a 3-22 minute one-way delay, so every Mars landing has been autonomous. See Crewed vs Uncrewed Requirements.
Crew
Crew Raise the Firm Load About Eightfold
The load that can never pause, at the same early lunar site.
| Load that cannot pause | Four crew | No crew | Rated |
|---|---|---|---|
| Life support | 15 kW | 0 | Modeled |
| Thermal control | 6 kW | 1-2 kW | Modeled |
| Avionics, communications, compute | 4 kW | 1.5-2 kW | Modeled |
| Firm load | 25 kW | About 3 kW | Modeled |
China
Chinese Programs on Every Substrate
Reactor, heater units, ice survey, regenerative life support and orbital compute.
| Program | Substrate | Data point | Rated |
|---|---|---|---|
| ILRS lunar nuclear power station · CNSA and Roscosmos | Energy | Delivery 2035, completion around 2036 | Projected |
| Yutu-2 · CNSA | Energy | Radioisotope heater units through every lunar night since 2019 | Observed |
| Chang'e-7 · CNSA | Water | Shadowed-crater ice survey; launch postponed 24 August 2026 | Building |
| Tiangong · CMSA | Water | Regenerative water recovery and electrolysis oxygen since 2022 | Observed |
| Three-Body Computing Constellation · ADA Space and Zhejiang Lab | Compute | 12 satellites, up to 744 TOPS each, flown May 2025 | Observed |
Contested
Deliberately Not Claimed
The load-bearing uncertainties, tagged with the confidence they actually earn.
0.37 t/kWe is one outpost's bill of materials. It assumes the lunar mix of solar, storage and a first reactor; a reactor-only Mars or Callisto site has a different factor, not yet derived.
Realized water energy is a range, not a measurement. No extraction plant has run on lunar ice; the 2.5-3.5 times multiplier on ideal heat is an engineering estimate.
Metals reduction has no plant mass. No off-Earth smelter mass has been derived anywhere in the analysis; the rung is excluded on its energy chain alone.
The storage crossover depends on tank technology. Cryogenic storage of hydrogen and oxygen at the lunar poles could cut fuel-cell tankage well below the high-pressure figure used here.
Record
Provenance and Data Sources
- Confidence
- Observed for flown, fielded or measured - Building for hardware and programs in development - Projected for announced plans - Modeled for original analysis
- Status date
- Q4 2026
- Analysis
- Closure Sequencing, Revision C (August 2026); The Primitives and its Machine Shop and Composite-Structure annexes; The Triple Substrate
- Physical data
- Solar constant 1,361 W/m2; Stefan-Boltzmann law; heat of formation of water; LCROSS Cabeus result (2009); NASA crew consumables
- China
- CNSA-Roscosmos agreement (May 2025); Yutu-2; Chang'e-7 postponement (24 August 2026); Tiangong; Three-Body Computing Constellation
- Weakest link
- Lunar ice grade and the 0.37 t/kWe factor outside the Moon