Lunar Industry

Lunar industry means treating the Moon as a production site rather than a destination: prospecting a resource, digging it, and converting it into a commodity — oxygen, water, propellant, metal feedstock — together with the power, thermal, transport and legal apparatus that makes any of that possible. The framing under test is that the Moon is the first industrial site off Earth. That framing is half true in a way worth spending a whole brief on, and the half that is true is not the half most readers expect.

This brief and Moon-Based Manufacturing divide at a unit-operation boundary, and both pages state it in the same words. This brief owns resource to commodity: finding the material, excavating it, beneficiating it, and the extraction chemistry that turns regolith into oxygen or ice into water — plus power, thermal, dust, delivered cost, the economics and the law. FR-II-23 owns commodity to object: forming, sintering, printing, casting, bagging and assembly, and whether making a thing on the Moon beats shipping the thing. Oxygen sits here because oxygen is a commodity, not an object.

The page is dense with numbers because, for the first time, there are numbers. Four commercial landers have flown and their outcomes are countable. A crewed free-return mission flew in April 2026. The delivery cost has been audited by the auditor of the agency paying it. The oxygen chemistry has a measured energy figure in grams per kilowatt-hour. And the resource question has a twenty-year-old radar result that constrains it more sharply than anything since. What has not happened is any industrial act: not one gram of anything has been extracted, refined or fabricated on the lunar surface by anybody, ever. Holding both halves at full strength is the whole job. For the 1970s programme history that generated the idea — the Ames summer study, the mass driver, the solar-power-satellite economics — see Historical Space Colonization Concepts; this brief starts where the measurements start.

Established Start with the tally, because the tally is the finding. The Commercial Lunar Payload Services record, as flown: Astrobotic's Peregrine Mission One launched 8 January 2024, suffered a propellant leak, never reached the Moon and re-entered on 18 January. Intuitive Machines' IM-1 Odysseus landed 22 February 2024 — the first US soft landing since 1972 — and tipped onto its side, with instruments partly operational. Firefly's Blue Ghost Mission 1 landed in Mare Crisium on 2 March 2025 and completed its objectives. Intuitive Machines' IM-2 Athena landed 6 March 2025 about 1,300 feet off target and on its side, and the mission ended roughly fifteen hours later. Established Two of four landed upright and functional, and both Intuitive Machines landers tipped over. The demonstrated success rate of simply arriving intact is about 50%, and that is the most repeated engineering outcome in the modern lunar record.

Established The tipping cost the resource question its only instrument. IM-2 carried PRIME-1 — the TRIDENT drill plus a mass spectrometer — the flagship attempt at the first direct subsurface volatile measurement at the lunar south pole. NASA's own account: the lander “was on its side, preventing it from fully operating the drill and other instruments”; the drill “successfully demonstrated the hardware's full range of motion”; NASA received 250 megabytes of data; and the mass spectrometer detected “elements likely due to the gases emitted from the lander's propulsion system.” It smelled its own exhaust. No subsurface ice measurement was obtained. Established Precision and attitude are separate problems and only one is solved: JAXA's SLIM landed within 100 m of its target in January 2024 — against Apollo 11's landing ellipse of roughly 20 km by 5 km — and then came to rest at about 90°, nose-down, having lost an engine nozzle on descent.

Established Crewed return is no longer a forecast. Artemis II launched 1 April 2026 from LC-39B, carried Wiseman, Glover, Koch and Hansen on a free-return trajectory, reached 406,771 km from Earth — past Apollo 13's record — and splashed down on 11 April after 9 days 1 hour 32 minutes. The anomalies are worth one clause because they are the texture of real operations: urine froze in the toilet vent lines on flight day 3, a small helium leak appeared in the European Service Module on flight day 8, and three of four CubeSats failed to raise their orbits. Established The slip record behind the launch is the other half: NET September 2025 as of January 2024, a December 2024 move to April 2026 over heat-shield char loss and life-support work, a hydrogen leak found in the 2 February 2026 wet dress rehearsal, and a helium flow issue on 21 February that returned the vehicle to the VAB.

Established The number that reorganises everything else is the audited price of a kilogram on the lunar surface. NASA's Office of Inspector General, IG-24-013 (June 2024), on CLPS: an indefinite-delivery contract with a maximum value of $2.6 billion through 2028; ten delivery task orders awarded, eight active and two terminated; an initial combined value of $781.4 million against a current value of $984.3 million, a 26% increase, of which $171.4 million — 82% of the total increase — came from NASA's own requirements changes; an average schedule delay of 14 months per task order; and a cost per kilogram delivered to the lunar surface of about $1 million in 2019 estimates and about $1.2 million in 2023 pricing, a 20% increase. Established The auditor's judgment on the contracting model is equally blunt: the initiative “does not yet meet the optimal conditions for success using FFP for service contracts... because of the immature market, changing lander requirements, inexperienced vendors, and financial and technical risks.” This is NASA's own audit arm reporting that NASA's flagship commercial lunar mechanism is getting more expensive per kilogram, that most of the overrun is NASA's fault, and that the market is not mature enough for the contract type NASA chose. Interest running hard against the finding, which is the strongest form an unwelcome number can take.

Established The campaign around it is priced by the same auditor and the verdict is one word. NASA OIG's February 2024 congressional testimony CT-24-001 gives $93 billion as the total projected Artemis campaign cost for FY2012–FY2025 and $4.1–4.2 billion per launch for a single SLS/Orion system through at least Artemis IV — excluding roughly $42 billion already spent on development. In the OIG's own words, continuing to rely on such an expensive heavy-lift system “will inhibit NASA's ability to reach its long-term human exploration goals in a timeframe anywhere close to its stated schedule”, and failure to reduce costs “will ultimately make the Artemis campaign unsustainable” under flat budgets. Frontier The lander leg carries its own unproven step: SpaceX “requires a series of Starship vehicles to establish a ‘fuel depot’ in low Earth orbit to refuel each Starship heading to the Moon,” with an uncrewed lunar landing demonstration mandatory before crewed Artemis III. The commonly cited figure of about ten tanker launches per mission is an estimate that has moved and that SpaceX has not firmly published.

Established Water splits into three propositions and only the first is settled. Proposition one: hydrogen and water are present. On 9 October 2009 a Centaur upper stage struck Cabeus; Colaprete and colleagues reported in Science near-infrared absorbance attributed to water vapour and ice and ultraviolet emission from hydroxyl, with 155 ± 12 kg of water vapour and ice in the instrument field of view at maximum and 5.6 ± 2.9% water ice by mass in the impacted regolith. Established Quote the second number with its error bar: ±2.9 on 5.6 is a 52% relative uncertainty, and a page that writes “5.6%” bare has thrown away the most important thing about it. Light hydrocarbons, sulphur-bearing compounds and carbon dioxide were also seen. Established Diviner puts subsurface temperatures in one of the coldest LCROSS-impacted regions at 38 K, with large polar areas cold enough to cold-trap water and a range of both more and less volatile species. Established Solar-wind water in the bulk regolith is real and dilute: Chang’E-5 samples give a soil average near 170 ppm H₂O, with individual minerals at 134–385 ppm and the top 200 nm of grain surfaces at 916–5,962 ppm. That is water. It is not a water resource at any plausible processing rate.

Frontier Proposition two: the ice is distributed in a form nobody has characterised. Hayne, Aharonson and Schörghofer revised the total cold-trap area upward to about 40,000 km², roughly 60% of it in the southern hemisphere, and found that 10–20% of the permanent water cold-trap area sits in micro cold traps — shadows on scales from 1 km down to 1 cm, “the most numerous cold traps on the Moon.” The authors read that optimistically: water “may be more widely distributed and accessible as a resource... than previously thought.” Frontier The pessimistic reading of the identical fact is equally available and this brief carries both. If much of the cold-trapped area sits in centimetre-to-metre shadows, the resource is not a deposit but a dispersion, and every kilogram recovered means moving a rover between shadows smaller than the rover. That is an inference from the published geometry rather than an argument Hayne and colleagues make, and it is flagged and attributed as such.

Established Proposition three — that the ice is minable — is not established, and there is direct evidence against thick deposits. Campbell, Campbell, Carter, Margot and Stacy, Nature 443 (2006), “No evidence for thick deposits of ice at the lunar south pole”: 20-metre-resolution, 13-cm-wavelength radar imaging found no evidence for concentrated deposits of water ice in Shackleton crater or elsewhere at the south pole. The circular-polarisation signatures previously read as ice appear at all latitudes and track rocky crater walls and ejecta rather than permanent shadow. If the Lunar Prospector hydrogen signal is water ice, the radar is “consistent with the ice being present only as disseminated grains in the lunar regolith.” Established This paper is twenty years old and remains the sharpest constraint in the subject, and it does not contradict LCROSS — it agrees with it. Disseminated ice at 5.6 wt% in regolith is precisely what the radar allows. Together the two results say the resource is dilute and mixed, not layered. Established NASA's own cost-benefit study concedes the point in its own premise: Jones and colleagues at NASA Langley and Analytical Mechanics Associates write that “the existence of such an architecture hinges on the abundance of accessible lunar ice, which has not yet been verified.” That is the paper arguing for lunar propellant, stating that its foundation is unverified. Frontier The instrument that would settle it has no firm ride: VIPER carries a neutron spectrometer sensitive to hydrogen to 1 m depth, a near-infrared spectrometer, a mass spectrometer and a 1-metre drill — the deepest planned for any robotic mission off Earth — over a mission of at least 100 days, and would establish where volatiles are, “how easy it is to access” them, and “how much of it is in the form of ice crystals versus being bound to minerals.” NASA cancelled it on 17 July 2024 at a stated project value of about $660 million, then on 19 September 2025 awarded Blue Origin a 2027 surface delivery conditional on performance milestones. Established State the epistemic position plainly: as of September 2026 there has never been a direct in-situ measurement of volatile abundance below the lunar surface at a polar cold trap. Everything about extractability is inference from orbit plus one impact plume.

Established Oxygen is the resource that needs no prospecting, and its feedstock claim is solid with a qualifier that is usually dropped. Lunar regolith is 40–45% oxygen by weight — chemically bound in mineral oxides, not available. Every gram costs energy. Frontier A KBR assessment for ESA scored eight extraction routes on yield, TRL, byproducts and confidence: hydrogen reduction about 1% yield at TRL 5; carbothermal reduction 20% at 1200 °C and 28% molten at 1600 °C, also TRL 5; molten regolith electrolysis about 16%; fluorination 32–44.8% conceptually and on terrestrial rock only; vapour-phase pyrolysis 1–23% with a realistic 1–8.5%. The ranking is molten salt electrolysis 51.72, carbothermal 49.01, hydrogen reduction 45.24, molten regolith electrolysis 42.30 — with the authors' own note that “all processes separated by only 8 points.” Established No route dominates, and that is a more honest statement than “the chemistry is demonstrated”. Established The best-instrumented demonstration supplies the energy number: NASA's Carbothermal Reduction Demonstration ran a brassboard reactor in the 15-foot Dirty Thermal Vacuum Chamber at Johnson, on SiO₂ + 2C to Si + 2CO with a methane loop and a Sabatier reactor downstream, reporting 13.42 g O₂ per kWh thermal at ambient and 11.53, 15.79 and 10.77 g/kWh in three vacuum runs. Frontier Derived, from two published inputs: at about 13 g O₂/kWh, one tonne of oxygen costs roughly 77 MWh of heat, and Crawford's independent figure for ilmenite reduction — 2–4 megawatt-years per 1,000 tonnes, i.e. 17.5–35 kWh/kg — is the same order and somewhat better. The arithmetic is ours and inherits the weaker flag. Either way, lunar oxygen costs tens of kilowatt-hours per kilogram. Frontier ESA's own programme states the timeline that did not happen: a molten-salt-electrolysis prototype at ESTEC operating at 950 °C in calcium chloride, with “the first technology demonstration targeted for the mid-2020s.” It is now September 2026 and no oxygen-extraction hardware has operated on the lunar surface. The interest runs toward the claim; the outturn runs against it.

Established Power is where the dependency on Advanced Nuclear Propulsion becomes a specific costed design rather than a gesture at “nuclear”. Oleson and colleagues at NASA Glenn specify a deployable 40 kWe lunar fission surface power unit for a 10-year life, at a total system mass of about 10,046 kg against a design goal of 6,000 kg — 67% over its own mass target. Reactor shielding alone is 1,250 kg of lithium hydride and tungsten inside a roughly 3,969 kg fission power subsystem; the main radiator is 133.4 m² at the south pole, rising to 216.2 m² at the equator, a 62% increase because equatorial heat rejection is harder. Deployment is three elements over two rover trips with the reactor placed 1 km from the base. NASA's own statement of why solar will not serve: “the sun is only available roughly two weeks out of every month — necessitating either a large energy storage system... or a nuclear reactor.” Frontier Now do the multiplication. At the audited $1.2 million per kilogram, a ten-tonne reactor is a $12 billion delivery charge before it is switched on. That single derived line explains more about why lunar industry has not started than any discussion of chemistry. Reactor physics, fuel cycle and propulsion applications belong to FR-I-11; this brief carries only the surface constraint.

Established Dust is the best-evidenced hazard in the slot and it has an operational record, not a model. Gaier's NASA technical memorandum on Apollo EVA systems catalogues nine categories — vision obscuration, false instrument readings, coating and contamination, loss of traction, clogging of mechanisms, abrasion, thermal control problems, seal failures, and inhalation and irritation — with mission-specific outturns. Apollo 12 suit pressure decay rose from 0 to 0.15 psi/min after the first EVA and to 0.25 psi/min after the second, against a 0.30 psi/min limit; Conrad's assessment was that “one or two additional EVAs could have resulted in a pressure failure.” All environmental and gas sample seals failed. Conrad reported the suits were “more worn after 8 hr of surface activity” than training suits after 100 hours. Apollo 16's gauge dials were scratched unreadable; the Apollo 12 magnetometer ran about 68 °F hotter than expected; total EVA time across six landings was 21–75 hours per mission. Cernan: “I think dust is probably one of our greatest inhibitors to a nominal operation on the Moon.” Established The health limit exists and is strict: LADTAG recommended a preliminary permissible exposure limit of 0.3 mg/m³ for a six-month lunar mission, derived from rat inhalation studies with authentic lunar samples and benchmarked against crystalline quartz at 0.1 and titanium dioxide at 5. Apollo symptoms were transient and resolved within 24 hours. Frontier And the caveat in the source's own words: “whether ground lunar material is representative of actual LD on the lunar surface, is at present unknown.” Mitigation is not flight-proven: NASA Glenn's own status is passive coatings and textured surfaces in ground testing, electrodynamic dust shields and lofting developmental, and “flight testing is the next step.”

Established And then the programmatic record, which is the quietly devastating part. The ISRU Capability Roadmap Final Report of 19 May 2005, chaired by Gerald Sanders of NASA JSC with Michael Duke of the Colorado School of Mines, set dated milestones: lunar regolith excavation demonstration 2010; Mars atmosphere collection demonstration 2011; lunar oxygen production pilot plant 2012; lunar polar water extraction demonstration 2012; Mars water extraction 2013; lunar oxygen production at pilot scale 2017; pilot-scale lunar construction 2017. Established Every dated lunar surface milestone was missed. The only two achieved are the Mars ones, and both were achieved by MOXIE about a decade late. The same document assigned lunar oxygen production TRL 3–4 in 2005; the 2024 KBR assessment puts hydrogen reduction and carbothermal reduction at TRL 5. Established State that as a rate: this is a field that has advanced about one technology-readiness level in two decades on its core process. Frontier And NASA's current plan does not lean on it. The 2024 Moon to Mars Architecture places “production of goods and services derived from lunar resources” in Sustained Lunar Evolution — the third of four segments — with Human Lunar Return and Foundational Exploration requiring no ISRU at all, and identifies 56 architecture-driven technology gaps. The agency running the return does not treat lunar industry as the purpose of the programme.

Established The dust record above concerns dust that is already on the ground; the harder problem is dust that a landing throws into the sky. Apollo 12 set down beside Surveyor 3 and the Surveyor parts the crew cut off and returned came back pitted and discoloured on the face that had looked toward the lunar module. Established That hardware is the only direct physical sample anyone has of rocket-driven regolith striking a spacecraft, and Immer, Metzger and colleagues analysed the veneer in Icarus as a calibration point for descent-engine erosion rather than as a curiosity. Frontier What turns it from a nuisance into a site-planning constraint is the absence of an atmosphere: ejected grains fly ballistically with no drag, so a particle leaving the surface at a few hundred metres per second lands tens of kilometres away, and the fastest grains in the published erosion models approach the 2.38 km/s escape velocity. Established Metzger and co-authors put the soil lifted during a single lunar module landing at of order a tonne.

Established Why the dust sticks is a two-mechanism answer and only one of the two is intuitive. Below roughly ten micrometres, van der Waals adhesion exceeds lunar weight by orders of magnitude, so the brush-and-shake that clears terrestrial sand does nothing; the second mechanism is electrostatic, and the Moon charges. Established Sunlit surfaces charge positive by photoemission at the scale of volts to tens of volts, while shadowed surfaces and the plasma wake charge negative, with shadowed potentials modelled into the kilovolt range during solar energetic particle events. Frontier Surveyor 7 photographed a horizon glow at the terminator that has been read since as forward scattering from electrostatically lofted grains, and the Stubbs, Vondrak and Farrell fountain model is the standard published account of the mechanism. Speculative The lofted population is the disputed part, and the disconfirming result is rarely carried in the mitigation literature: LADEE’s dust instrument found a permanent, asymmetric cloud consistent with meteoroid impact ejecta and did not find the dense persistent levitated layer the fountain model had been used to predict. Frontier For a plant designer the distinction is operational rather than academic: impact-generated dust arrives from above at random and is a shielding problem, charged surface dust arrives from the ground with every rover pass and is a removal problem, and the two buy different hardware.

Frontier The open questions in this subject are almost all measurements nobody has made, which is a very different epistemic situation from “no mechanism is known”. Nobody has drilled into a permanently shadowed region. Nobody has run an extraction reactor on the lunar surface. Nobody has operated machinery through a polar lunar night by design. That means most live positions here are decidable by an instrument rather than by a theory, and the hypothesis space should be read with that in mind.

Frontier Position one, and the one everything rests on: polar water ice is a minable resource at industrial scale. Held by NASA's ISRU programme, by the Sustained Lunar Evolution segment, by Metzger, and by most of the commercial lunar sector. For: LCROSS at 5.6 ± 2.9 wt% in Cabeus; about 40,000 km² of cold trap; ice-stability area roughly doubling at 2 cm depth. Against: Campbell's radar null on thick deposits, the 52% relative uncertainty on the LCROSS number, and NASA Langley's own concession that accessible ice has not been verified. VIPER's 1 m drill, or an equivalent in-situ measurement, would settle it, and this brief refuses to collapse the question further than frontier.

Frontier Position two is the most under-represented live position in the subject: the ice is present but as a dispersion rather than a deposit, which makes extraction a materials-handling problem rather than a mining problem. Both inputs are established — radar consistent only with disseminated grains, and 10–20% of cold-trap area in shadows down to 1 cm scale — but the conclusion is an inference this brief draws rather than a claim any fetched source makes in this form. Frontier If it is right, then the engineering problem is not a drill and a haul road; it is a rover that must process enormous volumes of cold, abrasive regolith at 38 K to concentrate a few percent of volatiles, which is a different machine with a different mass and a different power budget.

Frontier Position three: oxygen from regolith is the real lunar resource and water is a distraction. Held implicitly by ESA's molten-salt programme, the carbothermal community and the KBR ranking. It is stronger than the popular literature allows: regolith is 40–45% oxygen everywhere, so there is no prospecting risk, no permanently-shadowed traverse, no 38 K machinery, and two routes at TRL 5. Against it: tens of kWh per kilogram, and oxygen is only about 78% of a LOX/LH₂ propellant load by mass and useless without the fuel. Frontier Position four: the binding constraint is power, not chemistry. The Azami review states it outright — “a major impediment is the need for power” — and the arithmetic supports it: 13 g O₂/kWh thermal, 233 kWh to heat a tonne of regolith to 1000 °C, and a 40 kWe reactor massing ten tonnes against a six-tonne goal. The arithmetic is established; the causal claim about what actually stops the programme is frontier.

Frontier Position five is the rival causal claim and it is the one this brief finds hardest to argue against: the binding constraint is delivered cost per kilogram, and nothing else matters until it moves. Held implicitly by the OIG record itself. The number is $1.2M/kg, rising 20% in four years, on a programme explicitly designed to lower it. Established The number is established; the causal claim is frontier, and the honest way to hold positions four and five together is that power is what makes a plant expensive to run and delivery is what makes it expensive to have.

Frontier Position six: lunar propellant will be commercially viable and will lower the cost of doing everything else in space. Metzger argues it explicitly and the model is the most developed published case for lunar industry. Its terms: a gear ratio G of about 6–15 — kilograms that must leave Earth per kilogram of capital delivered to the lunar surface; a production mass ratio φ from 3.7 to 534 — kilograms of propellant produced per kilogram of capital, with “tent sublimation” topping the range near 500; a baseline launch cost falling from $2,000/kg to $30/kg over thirty years; parity with terrestrial propellant at GTO by year 6 and approaching LEO by year 30; and the competitiveness condition, minimise (x + G)/φ, which means maximising φ dominates everything. The counter-intuitive sensitivity is worth quoting: cheaper launch does not kill lunar propellant, because “terrestrial propellant price is exponentially dropping during this period, but lunar propellant drops faster.” Frontier What the model requires that is not in evidence is the crux. φ = 534 assumes a tent-sublimation scheme operating on ice that radar found no thick deposits of and that no instrument has measured in place; φ = 3.7 is the bottom of the same author's own range, a 144-fold spread in the decisive parameter. The entire economics of lunar industry turns on one number that spans two and a half orders of magnitude and whose true value depends on a subsurface measurement nobody has made. Frontier The independent NASA-side model agrees about the uncertainty: Jones and colleagues give plant mass of 0.133 to 1.110 tonnes per tonne per year of production — an 8.3-fold spread — and power of 0.617 to 5.698 kWe per tonne per year, a 9.2-fold spread, against modelled demand cases of 8.5, 17 and 34 t/yr on the surface. Their own caveats: “there has only been modest investment in ISRU systems over the last few decades, resulting in low technology readiness levels with significant uncertainty”, and, because the analysis assumes lossless transfer, zero boil-off and highly reliable autonomy, “an actual implementation of lunar ISRU will likely require greater cost than what is estimated here.”

Handwave Position seven is the one to debunk with arithmetic rather than with disdain: helium-3 is a lunar energy resource. Crawford's review gives solar-wind-implanted abundances from Apollo: hydrogen 46 ± 16 ppm (76 g/m³); helium-4 14.0 ± 11.3 ppm (23 g/m³); helium-3 0.0042 ± 0.0034 ppm (0.007 g/m³) — about 4 parts per billion by mass, with the highest measured concentrations near 10 ppb. Crawford, himself a lunar-science advocate arguing against the headline resource claims, writes that claims of abundance are “actually far from being the case”, that lunar ³He could “only ever make a relatively small contribution to Earth's total long-term energy needs”, and that the premise is “at least premature.” Speculative There is now a funded commercial claimant, which makes the multiplication worth doing openly. Interlune's own press release describes an excavator designed to ingest 100 metric tons of regolith per hour, with “several missions to the Moon later this decade” — and gives no helium-3 concentration, no recovery rate, no depth, no customer, no price and no dates. Speculative Derived, from two published inputs: at 4 ppb by mass, 100 t/hr yields 0.4 grams of ³He per hour; running continuously for a year, moving 876,000 tonnes of regolith, gives about 3.5 kg. Crawford supplies the concentration, Interlune the throughput, and the multiplication is ours. That is not a refutation of Interlune's business — small quantities of ³He are genuinely valuable for cryogenics and neutron detection — but it is a refutation of ³He as an energy-scale lunar export, which is the claim the popular literature actually makes.

Speculative Position eight: rare earths and metals justify lunar mining. Crawford is equally direct: maximum urKREEP total rare-earth content is about 1,200 ppm, and real crustal samples contain only a component of urKREEP, putting lunar rare earths “very much at the lower end of the scale for economically exploited REE deposits on Earth.” Frontier Position nine: the Moon is a proving ground for Mars and its industrial value is instrumental. As a description of NASA's actual plan this is established — ISRU appears in the Mars segment and the lunar segments do not require it. As a claim that the skills transfer it is frontier and probably overstated: lunar and Martian ISRU share almost no chemistry, since one is high-temperature reduction of silicate rock and the other is solid-oxide electrolysis of atmospheric CO₂. What transfers is operations, autonomy, dust handling and cost discipline.

Frontier Position ten: dust, not ice or power, is what will actually stop a sustained surface operation. Better evidenced than it is usually given credit for — seal decay to 83% of the limit in two EVAs, all sample seals failed, eight hours of wear exceeding a hundred training hours — and countered only by mitigation technologies that are in ground test with flight testing named as the next step. Speculative Position eleven: two competing legal blocs will fragment lunar resource rights. This is a prediction, not an observation: three states sit in both frameworks and nothing has been extracted, so nothing has been tested. Speculative Position twelve: a self-sustaining lunar industrial base. The settlement tradition's terminal claim. The useful thing a brief can do with it is say what it would take rather than when it will happen — and on the numbers above, what it would take is a resource measurement, a delivered cost two or three orders of magnitude lower, and a power system that does not cost twelve billion dollars to put down.

Frontier Position thirteen, which is missing from most of this literature: the binding dust constraint is not attrition of one machine but interference between machines, and it appears the first time two operators share a region. A landing is a sandblasting event with a range measured in kilometres and no air to stop it, so part of the cost of the second lander is paid by the first operator’s optics, radiators, seals and arrays. Established NASA has stood the problem up as a named discipline — plume-surface interaction — and has begun instrumenting it, flying stereo cameras on commercial landers to image crater formation and erosion during descent. Frontier That is the right instrument roughly five decades late: the erosion models from which every keep-out distance in the subject descends were calibrated against Apollo descent photography and subsonic jet experiments in terrestrial chambers. Speculative If the measured scaling is steeper than the models, the shared-site architecture every lunar economy assumes — landing zone, power plant, processing site and habitat within a few kilometres of one another — needs either prepared pads before the second landing or separations that defeat the point of sharing a site.

Established The first bottleneck is delivered cost per kilogram, and it is moving the wrong way. About $1.2 million per kilogram to the lunar surface in 2023 pricing, up 20% from the 2019 estimate, audited on a contract vehicle chosen specifically to bring it down. Cost to low Earth orbit has fallen roughly twentyfold since the Shuttle; cost to the lunar surface has risen over the life of the CLPS contract. Those two facts are not in tension — the lander is the expensive part, not the ride — and any argument that cheap launch will start lunar industry has to explain them.

Established The second is power, and it is a mass problem wearing a chemistry costume. At roughly 13 g of oxygen per kilowatt-hour thermal, or Crawford's 17.5–35 kWh per kilogram by a different route, oxygen production is a tens-of-kWh-per-kilogram business. The only credible continuous source is fission, the reference design masses about ten tonnes against a six-tonne goal, and delivering it costs about twelve billion dollars at the audited rate. Frontier The polar geometry makes this worse rather than better in one specific way: the power is on the ridges and the resource is in the holes.

Frontier The third is the unmeasured resource, and it is the only bottleneck that a single instrument could remove. Nobody has drilled a polar cold trap. VIPER was cancelled at about $660 million and re-manifested on a milestone-conditional 2027 delivery. Until that measurement exists, every economic model of lunar industry contains a parameter — Metzger's φ — that spans a factor of 144 and whose value is set by ground truth nobody has.

Established The fourth is dust, and it is the one with a real failure record rather than a projected one. Nine documented Apollo failure categories, suit pressure decay to 83% of limit within two EVAs, every environmental and gas sample seal failed, optics scratched unreadable, thermal control degraded by tens of degrees. A mining or extraction plant is a machine with seals, bearings, optical windows and radiators — every one of the Apollo failure modes. Mitigation exists in ground test and has never flown. Frontier The toxicological limit — 0.3 mg/m³ for a six-month mission — is derived from ground-milled material whose representativeness is, in the source's own words, unknown.

Frontier The fifth is arrival reliability, which nobody lists as a bottleneck and which the record puts at the top. Two of four CLPS landers arrived upright and functional; two Intuitive Machines vehicles tipped; SLIM landed within a hundred metres and then fell nose-down. An industrial site that has to receive a ten-tonne reactor in three pieces over two rover trips is a site that needs its landers to stay upright, and the demonstrated rate is about half. Frontier The sixth is autonomy, and it is the bottleneck with no number at all. Every architecture assumes surface hardware runs uncrewed for long periods; no source in this pack gives a mean time between interventions for any regolith-processing hardware, in vacuum, at temperature, in dust. That absence is worth naming as an absence rather than filling with an estimate.

Established And the seventh is that the entire ground evidence base is a proxy. NASA's own simulant guide states that agglutinates “have no terrestrial analog”, that nanophase iron droplets of 4–33 nm are “essentially impossible to synthetically create in quantities matching lunar samples”, that no simulant reproduces vacuum space-weathering, ion implantation or electrostatic charging, and that “no lunar simulant completely replicates all aspects of the lunar regolith.” The operative instruction is that “at least two very different lunar simulants should be used” even for a single technology. Moon-Based Manufacturing carries what that does to the fabrication numbers; the point here is that it applies equally to excavation forces, beneficiation yields and reactor feed behaviour.

Frontier The sixth bottleneck exists only once the programme starts working: plume-driven ejecta from neighbouring landings. Every constraint above is about one machine surviving its environment; this one is about a site surviving its own traffic, and it worsens in exact proportion to how well lunar industry goes. Frontier No published keep-out distance for a lunar surface asset rests on an in-situ measurement of ejecta flux at range, and the interim instrument — descent imaging of the crater a lander digs under itself — measures what leaves rather than what arrives. Speculative The cheap remedy is not technological: it is an agreed separation and approach geometry between operators, settled before the second operator arrives rather than after a solar array has been scoured, and no forum currently owns that decision.

Established The adjudication carries two edges and both are specific rather than thematic. The first is Advanced Nuclear Propulsion, which owns reactor physics, fuel cycle and the propulsion applications of fission. This brief waits on it for one concrete reason: lunar industry needs tens of kilowatts of continuous surface power and there is no credible alternative to fission. The NASA Glenn 40 kWe design is carried here because its mass, radiator area and deployment geometry are lunar surface constraints; everything upstream of the electrical output is FR-I-11's.

Frontier The second is Deep Space Infrastructure, which owns cislunar transport architecture, propellant depots, communications and navigation. This brief owns why you would want a depot and what the surface end of one looks like; the cislunar demand side, the transfer stages and the depot engineering are FR-I-25's. Metzger's propellant-market model straddles the two and is carried here as an argument about what the lunar surface can produce, with the demand side routed.

Frontier What this brief also depends on, and what no brief can supply, is a measurement. The dependency that would change the most is not on a technology at all: it is a metre-deep in-situ measurement of volatile abundance, form and distribution at a polar cold trap. VIPER was built to make it. It was cancelled, then re-manifested conditionally. No amount of progress in reactors, landers or extraction chemistry substitutes for that number, because the economics turn on it and the process selection turns on it — ice crystals and mineral-bound hydroxyl are different feedstocks requiring different machines.

Established What this brief does not depend on is more chemistry. Two oxygen routes sit at TRL 5, the comparative assessment finds all eight within eight points of each other, and the ESA and NASA demonstrations have both run. The gap between TRL 5 on simulant in a vacuum chamber and a plant on the surface is not a chemistry gap; it is mass, power, dust, autonomy and money, in that order.

Frontier The single most informative experiment available is a metre-class drill and a volatile-sensitive instrument suite inside a permanently shadowed region, operating for at least a hundred days. That is VIPER's specification and its purpose was exactly to answer the questions the economics turn on: where volatiles are, how easy they are to access, and how much is ice crystals versus mineral-bound. Established Nothing else in this brief would move as many numbers, and it is the one experiment that has been designed, funded, cancelled and re-manifested rather than merely proposed.

Frontier Second: an oxygen extraction pilot plant that runs on the surface rather than in a chamber. The 2005 roadmap put this in 2012 and the ESA programme targeted the mid-2020s; neither happened. What a surface run would establish that ground testing cannot is the interaction of a high-temperature process with real agglutinated, electrostatically charged, nanophase-iron-bearing feedstock in vacuum — every property NASA's own simulant guide says no simulant reproduces. Frontier CaRD's own roadmap named a lunar surface target of 2027; treat that as a schedule, not a result.

Established Third: fly the dust mitigation hardware. NASA Glenn's own status is that passive coatings and textured surfaces are in ground testing and electrodynamic shields are developmental, with the explicit statement that “flight testing is the next step to both verifying performance and tailoring ground based testing systems.” That is a stated gap with a named remedy, which is the most actionable kind.

Frontier Fourth: operate something through a polar night, by design, inside a shadowed region. The only thermal endurance data on the surface is accidental — SLIM was not designed to survive lunar night and survived four of them at about −120 °C without a radioisotope heater, setting the longevity record for a non-RTG lunar surface spacecraft — and Firefly's Blue Ghost managed 346 hours of lunar day and about five hours into the night. Nobody has operated anything inside a permanently shadowed region.

Frontier Fifth: measure the surface thermal environment properly, because the models are demonstrably wrong. Firefly's post-mission account is unusually candid: the surface was “hotter than expected and modeled”, the swings were “really, really crazy”, and reflected sunlight from an adjacent crater produced heating the models had missed, forcing the team to angle an antenna to shade its own radio. The LISTER heat-flow drill reached about three feet against a planned ten because it hit “really hard rock formations.” Both are first-order facts about operating machinery on this surface, both were unpredicted, and both went against the plan.

Established And sixth, a methodological requirement rather than a mission: run every ground result on at least two very different simulants. That is NASA's own instruction, it is routinely ignored in the published literature, and where it has been honoured the results are sobering — the same sintering process spans an order of magnitude in strength on simulant choice alone.

Frontier Seventh, and it does not displace the drill: measure the transfer function from simulant to regolith instead of assuming it. Qualifying hardware against a simulant yields a number whose relationship to the Moon is unmeasured, which is the defect this brief objects to everywhere else in the subject. Established The known differences are specific rather than vague: no simulant reproduces agglutinates, nanophase metallic iron, the fresh fractured-bond reactivity of grains that have never met an atmosphere, or the charging state of the real surface. Frontier The experiment is a coupon set — seal elastomer, bearing surface, thermal control paint, window glass, a cell cover — run to one fixed abrasion and adhesion protocol against two very different simulants on Earth and flown as an identical set on a lander with a simple exposure mechanism, with the ratio published rather than the lunar number alone.

Frontier The permanently shadowed regions have now been assessed as engineering sites rather than as scientific targets, and the result overturns the usual mental picture. Chen, Ye, Qiu and Li assessed 31 priority PSRs between 80°S and 90°S, defined as contiguous areas with summer maximum temperature below about 110 K, requiring candidate landing zones to have more than 40% average illumination over the 18.6-year precessional cycle and less than 5° slope. Their numbers: median traverse path lengths from a viable landing zone to the target of 12–50 km — best UN08 at 12.20 km, worst UN04 at 49.86 km and Cabeus2 at 46.32 km; 10–100% of the path exceeding a 25° slope threshold depending on target, with Shackleton and Stose exceeding 25° even on their best corridors while UN10 and UN11 stay below 15°; and minimum round-trip mechanical energy of 0.4–4.4 kWh. Frontier The surprising part is which way the numbers point. A few kilowatt-hours of traverse energy is nothing — the traverse is not the hard part. The hard parts are the slopes, the 12–50 km round trip into a place at 38–110 K with no sunlight at all, and the requirement that whatever goes in carries or beams its own power. And Shackleton — the crater with the most name recognition in the entire settlement literature — comes out Tier III, severely constrained. The authors' own summary: “geometric proximity does not equate to accessibility; in rugged terrain, safe corridors often require significant detours.”

Established Power deployment is a specified engineering sequence, not a concept. Three elements over two rover trips; reactor 1 km from the base for shielding standoff; controller 50 m from the reactor; cable spool at the user end; 1,250 kg of lithium hydride and tungsten shielding; 133.4 m² of radiator at the pole. Frontier The equatorial penalty is instructive: the same 40 kWe unit needs 216.2 m² of radiator at the equator, a 62% increase, because rejecting heat is harder where the ground is hot. Anyone arguing for an equatorial industrial site on the grounds of continuous sunlight should carry that number.

Frontier The polar illumination story is not the fourteen-day night, and conflating them is a common error. The fourteen-day night is an equatorial constraint. The polar constraint is different and in some ways worse: ridges with more than 40% illumination over the 18.6-year cycle exist, which is why every architecture points there, but the illuminated ridge and the cold trap are not the same place, and the gap between them is the 12–50 km of 25° slopes above. The power is near the resource but not in it, and closing that gap is a cable, a beam or a battery, each with a mass.

Established Thermal survivability at the surface is demonstrated only by accident and only outside the shadows. SLIM survived four lunar nights at roughly −120 °C without a radioisotope heater, having not been designed to survive one. Blue Ghost operated 346 hours of lunar day and about five hours into the night. Nobody has operated equipment through a lunar night by design at the poles, and nobody has operated anything at all inside a permanently shadowed region, where the ambient is 38–110 K and there is no sun to point at.

Frontier Excavation resistance is the mechanical property with the least data and the worst surprise. The only in-situ datum is LISTER on Blue Ghost, which reached about three feet against a planned ten in “really hard rock formations.” PlanetVac on the same lander collected, transferred and sorted regolith using pressurised nitrogen — the only in-situ regolith handling demonstration anyone has — and that is collection and transfer, not mining. Every excavation force, bucket-fill factor and specific energy in the design literature is a simulant number.

Frontier Dust on a radiator is a mass problem, and this brief’s own power numbers turn it into arithmetic. A thermal control surface works by holding a low solar absorptance against a high infrared emittance, and a thin patchy deposit of dark iron-bearing regolith degrades the first far faster than the second; the Apollo 12 magnetometer running tens of degrees hot is the flight evidence rather than a model. Established The 40 kWe reference reactor already carries 133.4 m² of polar radiator inside a system 67% over its own mass target. Speculative The same logic runs through every optical aperture in a processing plant — furnace viewports, separation sensors, navigation cameras — with the Apollo record unambiguous that scratched optics are permanent rather than cleanable, because the abrasive is harder than the window and the cleaning stroke is what does the damage.

Established The nearest neighbour is Moon-Based Manufacturing and the boundary between them is a unit operation, stated identically on both pages. This brief owns resource to commodity: prospecting, excavation, beneficiation, extraction chemistry, power, thermal, dust, transport, cost per kilogram, the economics and the law. FR-II-23 owns commodity to object: forming, sintering, printing, casting, bagging and assembly. The metals-from-electrolysis byproduct sits on the line and is split by verb — producing the metal is this brief's, forming it is FR-II-23's. The energy arithmetic appears in both pages for different purposes: here it prices a commodity, there it prices a fabrication step.

Established Historical Space Colonization Concepts owns the 1970s in full — the 1975 NASA Ames and Stanford summer study, O'Neill's papers and books, the L5 Society, Mass Driver 1, and the NRC and OTA reviews of solar power satellites that removed the economic rationale. The one place the seam is tempting is the mass driver, because the classical architecture ran on lunar material launched electromagnetically. The fact worth carrying here is one clause: a 1970s lunar-export concept existed, it belongs to FR-X-08, and there is no operating mass driver and no lunar launch infrastructure of any kind today. In the other direction, FR-X-08 should route here for what the Moon actually turned out to be like.

Frontier Asteroid Mining is the load-bearing comparison and it deserves one honest paragraph. Asteroid material sits at the bottom of a far shallower gravity well and has no fourteen-day night, no 25° crater walls and no 38 K traverse. Against that it has transit times measured in months to years, a launch cadence set by synodic windows, and an in-situ record thinner than the Moon's — which is saying something, since the Moon's amounts to four landers and one impact plume. FR-II-08 owns the asteroid side.

Frontier Space Resource Economies owns the general economics — property regimes, market formation, valuation, who the customer is. This brief owns the lunar-specific numbers: $1.2M/kg delivered, the gear ratio, φ, the helium-3 arithmetic. Where a general economic principle is needed, it is routed. Frontier O'Neill Cylinders is a downstream customer for lunar mass and states its demand in tonnes per square metre of shielding; this brief supplies the cost of getting to the lunar surface and notes that no capability exists for getting mass off it.

Established The most transferable institutional fact on this page is that the strongest evidence about Artemis's affordability comes from inside NASA. The Office of Inspector General audits NASA, is funded by NASA's appropriation, and published that the flagship architecture costs $4.2 billion per launch excluding $42 billion of development, that the campaign will be unsustainable under flat budgets without cost reduction, and that the commercial delivery mechanism is 26% over its initial task-order value with 82% of the increase attributable to NASA's own requirements changes. Interest running against the finding, and it should be weighted higher than any external critique for exactly that reason.

Established The two-bloc framing is popular and the facts are more interesting than it allows. The Artemis Accords reached 70 signatories as of 17 July 2026, eleven of them joining during 2026 — Portugal, Oman, Latvia, Jordan, Morocco, Malta, Ireland, Paraguay, Botswana, Serbia and Mauritius. The International Lunar Research Station, founded by China and Russia in June 2021, stands at 13. Established Three states — Senegal, Serbia and Thailand — are in both frameworks, which is a better fact than “two rival blocs” and should replace it. Note also that details of the ILRS principles are not publicly available, so the comparison is between a published text and an unpublished one.

Established The legal position is precise and unresolved in a specific way. Outer Space Treaty Article II holds that outer space, including the Moon and other celestial bodies, “is not subject to national appropriation” by claim of sovereignty, use, occupation or any other means, and is treated as customary international law binding all states. Against that stand four domestic statutes: the US Commercial Space Launch Competitiveness Act (2015), under which US citizens may “possess, own, transport, use and sell” space resources obtained while the US does “not thereby assert sovereignty”; Luxembourg's Law on Space Resources (2017), stating that “space resources are capable of being appropriated”; the UAE's Federal Law No. 12 (2019); and Japan's space resources act (2022). China and Russia have objected to these statutes as inconsistent with non-appropriation. The Moon Agreement, which reiterates non-appropriation in its Article 11, has among Artemis signatories only Australia, Mexico and Saudi Arabia. Frontier The honest summary: there is no settled law on extracting and selling lunar material, the practice is being established unilaterally by domestic statute, and the two largest lunar programmes disagree about whether that is lawful. No dispute has arisen because nothing has been extracted.

Frontier The contracting model is itself an institutional finding. Firm-fixed-price service contracts work where the market is mature, the requirement is stable and the vendors are experienced. The auditor's judgment is that CLPS has none of those three. The observable outturn is a 50% arrival-intact rate, a 14-month average delay per task order, and a per-kilogram price that rose while the mechanism was supposed to lower it. Frontier That is not an argument against commercial procurement; it is an argument that the procurement instrument was chosen ahead of the market it assumed.

Speculative And an institutional pattern the roadmap record makes visible. The 2005 ISRU roadmap set seven capability elements and dated milestones, was signed by the agency's own subject-matter leadership, and missed every lunar surface date it set. The field's readiness moved from TRL 3–4 to TRL 5 in twenty-one years. A roadmap that is never audited against outturn is a planning document, not a commitment, and the most useful institutional reform this subject could adopt is the one its own Inspector General already practises: publish the variance.

Established The clearest ethical constraint is an occupational health limit that already exists. LADTAG's preliminary permissible exposure limit of 0.3 mg/m³ for a six-month lunar mission sits between crystalline quartz at 0.1 and titanium dioxide at 5, and it was derived from rat inhalation studies using authentic lunar samples rather than from analogy. Frontier The honest caveat is the source's own: whether ground lunar material represents actual dust on the surface “is at present unknown”. A workforce standard resting on a proxy of unknown fidelity is a governance problem before it is a scientific one, and it will bind the first sustained surface operation rather than the first sortie.

Frontier A less discussed conflict is between the resource and the science. The permanently shadowed regions are simultaneously the only candidate ore bodies and the best-preserved volatile archives in the inner solar system — cold traps at 38 K that may hold a record of volatile delivery over billions of years. Any extraction campaign destroys the sample it is mining. Speculative No framework currently allocates that trade-off, and the domestic resource statutes above are silent on it.

Frontier Third, the distributional question the legal section implies. Four states have legislated a private right to sell what is extracted from a body that a treaty says is not subject to national appropriation. The states with lunar delivery capability are a small set; the states that are parties to the Outer Space Treaty are almost all of them. Whatever one thinks of the legal argument, the practice is being written by the parties who can reach the surface, and that is a fact about power rather than about law. Space Resource Economies owns the analysis.

Speculative Fourth, and immediate rather than distant: investment claims. The helium-3 arithmetic in section 3 is not an argument against a company; it is an argument about a number. A published throughput of 100 tonnes of regolith per hour against a published concentration of 4 parts per billion is 3.5 kilograms a year, and a reader who does not do that multiplication will read “energy resource” where the geochemistry supports “speciality cryogen”. Established The appropriate response is exactly the one this page performs — state the vendor claim, state the measured concentration, do the arithmetic openly, and label it derived.

Frontier If lunar propellant works, the consequence is not lunar — it is cislunar. Metzger's model makes the strongest available version of the case: minimise (x + G)/φ, and lunar propellant reaches parity at GTO within about six years and approaches parity at LEO within thirty, lowering “the cost of doing everything else in space.” The gear ratio of 6–15 is the leverage: a kilogram of capital on the lunar surface costs six to fifteen kilograms leaving Earth, and after that a plant that makes hundreds of times its own mass in propellant changes the shape of every mission budget beyond LEO. Frontier The whole civilizational case is that one parameter, φ, and it spans a factor of 144 in its own author's paper.

Frontier The second-order effect is that a working lunar industry would change what a habitat costs, not just what a mission costs. O'Neill Cylinders states its demand in tonnes of shielding per square metre, and the classical answer to that demand was lunar material. The honest position from this side of the trade is discouraging: CLPS delivers to the surface at $1.2M/kg, nothing has ever been launched off the lunar surface except Apollo and Chang’e samples, and no beneficiation, no mass driver and no lunar launch infrastructure exists. The delivery cost is a cost of arriving, not of exporting, and the export number has never been measured because there has never been an export.

Established What is genuinely new, and worth saying warmly, is that this subject has stopped being an estimate. Ten years ago every number in this brief was a projection. Now the delivery cost is audited, the arrival reliability is countable, the oxygen chemistry has a measured yield in grams per kilowatt-hour, the reactor has a mass and a radiator area, the dust has a permissible exposure limit, and the traverse into a named crater has a distance and a slope distribution. The Moon is the first place off Earth where an industrial site is being seriously attempted, and the attempt is generating real cost, failure-rate and energy data rather than estimates. That is a different situation from 2020 and the page should celebrate it.

Speculative And the terminal claim, stated as a requirement list rather than a date. A self-sustaining lunar industrial base would need: a verified accessible volatile resource; a delivered cost per kilogram two to three orders of magnitude below today's; a power system whose delivery charge is not measured in billions; autonomy with a demonstrated mean time between interventions; and a legal basis for selling the product that the two largest lunar programmes agree on. None of those five is in hand, and only one of them — the resource measurement — could be settled by a single mission.

These horizons track measurements and audited costs rather than technology-readiness rhetoric, because this subject's record of dated self-prediction is uniformly bad — every lunar surface milestone in the 2005 ISRU roadmap was missed:

• 10 yr: Frontier Expect the resource question to be answered or to remain conspicuously open. VIPER or an equivalent metre-class drill delivered to a polar cold trap is the single event that would move the most numbers; its current status is a milestone-conditional 2027 delivery. Frontier Expect a surface oxygen demonstration at pilot scale — CaRD's own roadmap names 2027 — and treat that as a schedule rather than a result, given that the 2005 roadmap put a lunar oxygen pilot plant in 2012 and ESA targeted the mid-2020s. Established Expect fission surface power hardware to be built and tested on Earth; the 40 kWe design exists, masses about ten tonnes against a six-tonne goal, and costs about $12 billion to deliver at the audited rate, so expect the mass target to be the contested item. Speculative Do not expect the delivered cost per kilogram to fall by an order of magnitude on this horizon; it rose 20% over the last four years of the contract designed to lower it.
• 25 yr: Speculative If the resource measurement comes back favourable, this is the horizon on which a propellant plant could plausibly operate at the few-tonnes-per-year scale that Jones and colleagues model as the smallest useful demand case. Speculative If it comes back unfavourable — disseminated grains at a few weight percent across shadows measured in centimetres — expect the field to pivot to regolith oxygen, which needs no prospecting and is already at TRL 5, and to accept that hydrogen must be imported. That pivot is not a failure; it is a different industry with a different product. Frontier Expect the legal question to be tested for the first time, because a dispute requires an extraction and there has not been one.
• 50 yr: Speculative A sustained industrial presence at this horizon requires the delivered-cost curve to bend, and the only mechanism in evidence that could bend it is producing the propellant locally — which is circular, and Metzger's model is precisely an argument that the circle closes. Whether it closes depends on φ. Speculative Lunar export — material leaving the surface for use elsewhere — is the capability with the least evidence of any in this brief. There is no mass driver, no launch infrastructure and no demonstrated beneficiation, and the classical 1970s answer to this problem belongs to Historical Space Colonization Concepts rather than to any current programme.
• 100 / 250+ yr: Handwave Beyond useful forecasting, and the defensible statement is about rates rather than dates: lunar oxygen production moved from TRL 3–4 to TRL 5 in twenty-one years, and every dated surface milestone the field set for itself between 2010 and 2017 was missed. Speculative A subject advancing at roughly one technology-readiness level per decade, on a delivery cost that has been rising, has no base rate at this horizon. What can be said is that nothing in the physics or chemistry forbids a mature lunar industry — the obstacles are all mass, energy, money and one unmade measurement.

Frontier One dust milestone belongs on the ten-year line and is not in the list above: either an active removal system has operated on the surface and returned a performance number, and the electrodynamic shield manifested on a commercial lander in 2025 is reported to have cleared regolith from test surfaces — a result flagged rather than asserted here, because it post-dates the agency status page cited above and was not re-read or mitigation remains a ground-test literature with a named next step it has not taken.

• Depends on Two edges, both specific. Advanced Nuclear Propulsion because lunar industry needs tens of kilowatts of continuous surface power and fission is the only credible source: NASA Glenn's reference 40 kWe unit is specified for a ten-year life at about 10,046 kg against a 6,000 kg design goal, with 1,250 kg of lithium-hydride and tungsten shielding and 133.4 m² of polar radiator, and NASA's own justification is that “the sun is only available roughly two weeks out of every month.” This brief carries the surface constraint; FR-I-11 owns the reactor physics and fuel cycle. And Deep Space Infrastructure because the product of a lunar propellant plant is only worth making if there is a cislunar transport architecture to consume it — depots, transfer stages, navigation and communications are FR-I-25's, and Metzger's parity-at-GTO result is a statement about that architecture as much as about the Moon. The dependency that no brief can supply is a measurement: a metre-deep in-situ determination of volatile abundance and form at a polar cold trap, which VIPER was built for, cancelled at about $660 million in July 2024, and re-manifested on a milestone-conditional 2027 delivery.
• Requires (not on this map) The first is a measurement, not a technology: no direct in-situ determination of volatile abundance, form or distribution below the surface at a polar cold trap has ever been made, and the economics of the whole subject turn on a production mass ratio that spans a factor of 144 because of it. The second is that four states have legislated a private right to sell space resources, two of the largest space powers say those statutes are inconsistent with Outer Space Treaty Article II, and nothing has been extracted, so the disagreement has never been tested. A third is added by this pass and is administrative rather than scientific: agreed ejecta keep-out distances and approach geometries between adjacent surface operators. A landing is a kilometre-scale sandblasting event in a vacuum, the only direct measurement of the effect is the Surveyor 3 hardware Apollo 12 returned from a far lighter vehicle, and no forum owns the decision. It is a constraint rather than a concern because it binds the first time two operators share a region, which is the configuration every lunar economy described here assumes.
• Enables No typed enabling edge is claimed, and the reason is the record rather than modesty. Nothing has been extracted, refined or fabricated on the lunar surface by anybody, and NASA's own 2024 architecture places resource-derived goods in the third of four segments while requiring none of it for the first two. Moon-Based Manufacturing takes this brief's commodities as its feedstock and Mars Colonization depends on this brief for operations, autonomy, dust handling and cost discipline — but those edges are declared from the far side, where the depending brief can state what it is waiting for, rather than asserted from here as a claim about what lunar industry will unlock.
• Adjacent Moon-Based Manufacturing across the resource-to-commodity versus commodity-to-object boundary; Asteroid Mining as the shallower-gravity-well alternative with a worse transit time and a thinner in-situ record; Space Resource Economies for markets, property regimes and valuation; O'Neill Cylinders as the downstream customer for lunar mass, which this brief serves badly because no export capability exists; and Historical Space Colonization Concepts for the 1970s mass-driver architecture that first proposed all of it.

Handwave “Helium-3 is the reason to go to the Moon.” The measured concentration from Apollo samples is 0.0042 ± 0.0034 ppm — about 4 parts per billion by mass, with the highest measured values near 10 ppb. Crawford, a lunar-science advocate writing in a peer-reviewed review, states that claims of abundance are “actually far from being the case” and that lunar ³He could “only ever make a relatively small contribution to Earth's total long-term energy needs.” Speculative Derived, from a vendor's own throughput figure and a peer-reviewed concentration: Interlune's excavator is designed to ingest 100 tonnes of regolith per hour; at 4 ppb that is 0.4 grams of ³He per hour, or about 3.5 kilograms in a year of continuous operation. Speculative That is a real business in cryogenics and neutron detection and it is not an energy resource, and the difference between those two sentences is three orders of magnitude of demand.

Speculative “The Moon has rare earths worth mining.” Maximum urKREEP total rare-earth content is about 1,200 ppm, and real crustal samples contain only a component of urKREEP. Crawford's assessment is that this puts lunar rare earths “very much at the lower end of the scale for economically exploited REE deposits on Earth” — before adding a $1.2 million per kilogram delivery charge in the opposite direction.

Established “There is water ice at the lunar poles, so there is lunar propellant.” Three separate propositions are being run together and only the first is settled. Water is present: LCROSS measured 155 ± 12 kg in the plume and 5.6 ± 2.9 wt% in the impacted regolith. Its form and distribution are uncharacterised: 10–20% of the cold-trap area is in shadows down to 1 cm scale. And minability has evidence against it: 13-cm radar at 20 m resolution found no concentrated ice at Shackleton or anywhere at the south pole, with polarisation signatures tracking rocky walls and ejecta at all latitudes, consistent only with disseminated grains. Established NASA's own cost-benefit paper for lunar propellant says the accessible ice “has not yet been verified”. The instrument sent to close the question landed on its side.

Established “LCROSS found 5.6% water ice.” LCROSS found 5.6 ± 2.9 weight percent — a 52% relative uncertainty — in one crater, from one impact plume, inferred from the estimated mass of excavated regolith that reached sunlight. Dropping the error bar converts a measurement into a design assumption, and the design assumption is the thing every economic model of lunar industry is built on.

Established “The fourteen-day night is the problem at the poles.” It is the equatorial problem. The polar problem is that ridges with more than 40% illumination over the 18.6-year cycle exist and are not where the resource is: median traverses of 12–50 km into targets at 38–110 K, with 10–100% of the path above a 25° slope. Frontier And Shackleton, the crater with the most name recognition in the settlement literature, ranks Tier III — severely constrained — in the only high-resolution accessibility assessment of the 31 priority PSRs.

Established “Cheap launch will start lunar industry.” Cost to low Earth orbit has fallen roughly twentyfold since the Shuttle era. Over the same period, the audited cost per kilogram to the lunar surface went up 20%, from about $1 million in 2019 estimates to about $1.2 million in 2023 pricing, on the specific contract vehicle NASA created to bring it down. Frontier Metzger's model actually argues the same point from the other direction: cheaper launch does not remove the lunar advantage because lunar propellant prices fall faster — but nothing in the model says cheap launch alone starts anything.

Frontier “Artemis is a lunar industry programme.” NASA's own 2024 Moon to Mars Architecture places “production of goods and services derived from lunar resources” in Sustained Lunar Evolution, the third of four segments, and requires no ISRU for Human Lunar Return or Foundational Exploration. Established The agency running the return does not treat industry as the purpose of it, and the campaign's own auditor reports $4.2 billion per launch and a configuration it calls unsustainable under flat budgets.

Established “The extraction chemistry is demonstrated.” Two routes sit at TRL 5 on simulant, and the comparative assessment ranks eight routes within eight points of each other, meaning no route dominates and selection is mission-dependent. Established NASA's 2005 roadmap set a lunar oxygen pilot plant for 2012 and polar water extraction for 2012; ESA targeted a surface demonstration for the mid-2020s. Nothing has run on the lunar surface. The movement in twenty-one years is TRL 3–4 to TRL 5.

Speculative “Simulant results tell us what will happen on the Moon.” NASA's own simulant guide says agglutinates “have no terrestrial analog”, that nanophase iron of 4–33 nm is “essentially impossible to synthetically create in quantities matching lunar samples”, that vacuum space-weathering, ion implantation and electrostatic charging are not reproduced, and that “at least two very different lunar simulants should be used” for any single technology. Frontier The honest reading is that simulant work establishes process plausibility, not performance — and the one in-situ mechanical datum anyone has, LISTER's drill reaching about three feet against a planned ten, went the wrong way.

Speculative “Two rival legal blocs are carving up the Moon.” Seventy Artemis Accords signatories against thirteen ILRS partners, with Senegal, Serbia and Thailand in both, and no dispute of any kind because nothing has been extracted. Frontier What is real is narrower and harder: Article II says celestial bodies are not subject to national appropriation, four states have legislated a private right to possess and sell what is extracted, and China and Russia say those statutes are inconsistent with the treaty. That disagreement has never been tested and will not be until somebody sells something.

Established And the framing itself. “The Moon is the first industrial site off Earth” fails on the word industrial and survives on the word first. Established Nothing has been produced there: not one gram, ever, by anybody — and the only in-situ resource utilisation ever demonstrated in the solar system ran on Mars and made 122 grams of oxygen. Established The honest version is narrower and still worth a brief: the Moon is the first place off Earth where an industrial site is being seriously attempted, and the attempt is now producing audited costs, measured failure rates and published energy yields instead of estimates. The industrial claim rests on one unmeasured variable and one unsolved economic one, and this page names both rather than averaging over them.