Moon-Based Manufacturing

Moon-based manufacturing means turning lunar feedstock into objects: sintering, melting, casting, printing, bagging and assembling, on the surface, at one-sixth gravity, in vacuum, with an abrasive and electrostatically active raw material. The framing under test is that manufacturing on the Moon is the point of going back.

This brief and Lunar Industry divide at a unit-operation boundary, and both pages state it in the same words. FR-II-01 owns resource to commodity: finding the material, digging it, and turning it into oxygen, water, propellant or metal feedstock — plus power, dust, transport cost and the law. This brief owns commodity to object: forming, sintering, printing, melting, casting, bagging and assembling — everything that happens after you have feedstock, and the question of whether making an object on the Moon beats shipping the object. On that rule the ice question and the extraction chemistry belong next door and are cited once; the metals-from-electrolysis byproduct sits on the line and is split by verb, with producing the metal over there and forming it here.

Two further seams. Space-Based Manufacturing owns microgravity manufacturing — in-space assembly, additive manufacturing on orbit, fibre drawing, the ISS materials record — and the physics differs from this brief's in three specific ways: powder handling behaves differently under gravity than in free fall, thermal rejection differs on a regolith-conducting surface versus a radiating structure, and lunar feedstock is a mineral mixture rather than a specified alloy. And Historical Space Colonization Concepts owns the 1970s: the idea that lunar material would be processed rather than shipped is fifty years old and its 1970s form belongs there. What is new is that there are now measured compressive strengths, comparatively ranked routes and an audited delivery cost.

Established The strongest sentence available to this brief is a negative one and it belongs first: no object has ever been fabricated from lunar material, on the Moon or on Earth. Apollo and Luna returned 382 kg, Chang’e-5 and Chang’e-6 returned more, and none of it has been sintered, printed, cast or formed into a structural article except in microgram-scale laboratory experiments on returned sample material. Every published compressive strength in the literature below is a simulant number. Established The nearest thing to a lunar-material handling demonstration on the surface is PlanetVac on Blue Ghost Mission 1 in March 2025, which “collected, transferred, and sorted lunar regolith from the moon using pressurized nitrogen gas” — collection and transfer, not fabrication. Established And the nearest thing to subsurface access is LISTER on the same lander, which reached about three feet against a planned ten after hitting “really hard rock formations”. That is the only in-situ mechanical-property datum the field has, and it went the wrong way by a factor of about three.

Frontier The forming routes are now surveyed with numbers, which the popular treatment cites without. Azami, Kazemi, Moazen, Dubé, Potvin and Skonieczny's comprehensive review reports: laser powder bed fusion at 4.2–31.4 MPa for regolith alone and 264 MPa for a regolith–aluminium composite, at 1100–1300 °C with scan speeds down to 1 mm/s; solar sintering at 2.49 MPa at ambient pressure, with vacuum results not reported, up to 1350 °C and limited by “low precision of its beam focus”; microwave sintering, which penetrates to 65 cm and uses about 23% less energy than laser sintering while receiving a “relatively lack of attention”; and large-scale laser processing at 56–216 MPa, mean 93.97 MPa, using a 3 kW CO₂ laser at 5 mm/min — roughly fifty times the strength of solar-sintered specimens. Frontier And the review's own showstopper, quoted rather than paraphrased: “A major impediment is the need for power. The primary current source of power on the Moon is solar energy” — and, on the laser routes specifically, “the conversion of this energy into electricity for laser operations presents certain challenges, making laser-based additive manufacturing highly inefficient in the Lunar setting.”

Frontier The single most useful source for this slot is the one that puts the routes on a common scale, and its top-ranked answer is not manufacturing. Bao, Zhang, Wang, Cui and Feng, in Engineering (2024), score four families of regolith solidification against each other on in-situ material ratio, time, temperature requirement and equipment. The result: regolith bagging — confinement formation — ranks first at 3.80 out of 5, with 2–3 MPa compressive strength and a 99% in-situ material ratio. Casting comes second at the highest measured strength of 538 MPa and above 98% in-situ; solar melting third; direct sintering fourth; and microwave sintering fifth at 12–120 MPa depending on which simulant is used. Reaction and bonding solidification sit at 65–95% in-situ material. Established The mass penalty for the winning technique is small and stated: about 0.7 tonnes of external material for one inhabited module — 0.5 t of bags plus 0.2 t of airbags — roughly 1% imported and 99% in-situ by mass.

Frontier Why the ranking matters more than it looks, and this is the finding the brief is built on. The dominant near-term structural requirement on the Moon is radiation and micrometeoroid shielding, and shielding is a mass-per-unit-area requirement, not a strength requirement: O'Neill Cylinders gives the areal densities as 6–7 tonnes per square metre of polyethylene or 10–11 tonnes per square metre of lunar regolith for a 20 mSv/yr settlement dose target, and NASA Ames gives galactic cosmic radiation halved at 8 g/cm² and quartered at 50 g/cm². Frontier A material that is 2–3 MPa and 99% local satisfies the actual requirement. You do not need to manufacture anything to make a shielded lunar habitat; you need bags and a shovel. Manufacturing becomes necessary for pressure vessels, mechanisms, spares and electronics — precisely the items where strength and precision requirements are highest and lunar feedstock is worst suited. Frontier The counter to the counter should also be carried. Landing pads and roads are the one near-term construction need bagging does not meet, because the requirement there is a sintered or melted continuous surface to stop plume ejecta, and that is a real capability at a real energy cost: large-scale laser processing gives 56–216 MPa at 5 mm/min with a 3 kW laser.

Established Then the energy arithmetic, which converts an abstract bottleneck into a bounded number a reader can argue with. Bao and colleagues state that “heating 1 t of regolith to 1000 °C requires at least 233 kW·h” before heat losses, and put it in physical terms as roughly a 100 m² photovoltaic array running for six hours of lunar daylight. Frontier Derived, from that figure and NASA Glenn's reference power unit: a 40 kWe reactor running continuously produces 350 MWh a year; at 233 kWh per tonne and zero losses that is a theoretical ceiling of about 1,500 tonnes of regolith heated to 1000 °C per year — and 1000 °C is below the 1100–1350 °C the sintering routes actually use, while heat losses from an open process in vacuum are not zero. The realistic figure is some fraction of that. Both inputs are agency or peer-reviewed publications; the multiplication is ours and inherits the weaker flag. Frontier The same reactor's oxygen ceiling, for cross-comparison with Lunar Industry: at the Carbothermal Reduction Demonstration's measured 13 g O₂ per kWh thermal, 350 MWh a year gives about 4.6 tonnes of oxygen per year — against NASA Langley's modelled lunar-surface demand cases of 8.5, 17 and 34 tonnes per year. A single 40 kWe reactor does not reach even the smallest of them. That is the most concrete available statement of the scale of power a lunar industrial base implies. Frontier And the honest counter, which is the substantive finding again: microwave sintering uses about 23% less energy than laser sintering and penetrates 65 cm, and regolith bagging needs no high-temperature step at all. The energy wall is real for the routes that make strong material and largely absent for the route that makes shielding.

Established Every number above is a proxy number, and NASA has published the definitive statement of how weak the proxy is. The Lunar Regolith Simulant User's Guide of October 2024 lists what simulants cannot reproduce. Agglutinates — impact-welded aggregates with “very complex geometries” that “have no terrestrial analog”, with producers having “not achieved satisfactory results” reproducing their geometry at scale. Nanophase iron — metallic droplets of 4–33 nm formed in lunar glass by micrometeoroid impact, “essentially impossible to synthetically create in quantities matching lunar samples”. Space weathering — surface patinas, ion implantation, atomic displacement, and electrostatic charging in vacuum and plasma, none of it reproduced. And volatiles — solar-wind-implanted hydrogen and helium, absent from terrestrial simulants. The guide's blanket statement is that “no lunar simulant completely replicates all aspects of the lunar regolith”. Established Fidelity is scored by a Figure of Merit from 0 to 100 across eight properties — mineralogy, oxide chemistry, particle size distribution, particle geometry in aspect ratio and form factor, material density, magnetic susceptibility, and shear strength in cohesion and internal friction angle — and the guide flags the framework's own limits: data from multiple sources with varying instruments, small samples standing for bulk materials, and sampling bias. Four classes are defined: basic (BP-1, Greenspar, bunker sand — single rock type, no glass, low-readiness concept work); general purpose (about 15% glass, mixed rock types — most commercial and NASA work); enhanced (the NU-LHT series, with NUW-LHT-5M at 60% crystalline and 40% synthetic glass matching Apollo 16 chemistry); and specialty (ice-bearing regolith, accurate agglutinates with nanophase iron). Established And the operative warning, verbatim: “No single lunar regolith simulant will satisfy all the technology needs of a broad-based lunar exploration program. Even when developing and testing a single technology at least two very different lunar simulants should be used to determine any simulant-dependent results.” Frontier What that does to section 2's numbers is measurable rather than rhetorical. Bao and colleagues report microwave-sintered strength as 12–120 MPa depending on regolith simulant properties — a tenfold spread attributable to the choice of proxy rather than to the process. Every single-simulant compressive strength in the literature should be read as carrying an unstated order-of-magnitude uncertainty band. The same weakness appears in the biology: the lunar dust toxicity review states plainly that “whether ground lunar material is representative of actual LD on the lunar surface is at present unknown”. The proxy problem is not confined to manufacturing; it is the field's structural limitation.

Speculative The vendor claims should be assessed on their content rather than repeated, and a defensible grading rule can be stated openly. Blue Origin's Blue Alchemist is the most-cited example of “we can make solar cells on the Moon”. The company's own announcement claims iron, silicon, glasses and ceramics from a molten-regolith-electrolysis reactor, plus solar cells, power transmission wire, oxygen, metals and semiconductors; landings “up to 60% cheaper”; fuel-cell and battery mass reduced “up to 70%”; a 3-acre, 60,000 square foot Space Resources Center with 65 or more experts; a Critical Design Review completed in September 2025; and an autonomous demonstration in a simulated lunar environment planned for 2026. It states no silicon purity, no cell efficiency, no oxygen yield, no reactor temperature, and does not disclose the NASA Tipping Point award value. Frontier The independent counterpart is the KBR assessment for ESA, which puts molten regolith electrolysis at about 16% oxygen yield and fourth of eight overall at a score of 42.30, noting it has been “demonstrated by NASA/Blue Origin”. So the electrolysis is real and the solar-cell claim built on top of it is where the evidence thins. Frontier ICON's Project Olympus is the construction counterpart: a roughly $57.2 million NASA SBIR Phase III award announced on 29 November 2022 to develop a lunar surface construction system for landing pads, blast shields, roads and habitats, building on earlier NASA and Department of Defense funding and a 2021 Jacobs subcontract that produced Mars Dune Alpha at Johnson. The announcement states no technology readiness level and no target date for a lunar demonstration. Frontier The rule, stated so a reader can apply it themselves: a vendor capability claim with no performance number in it is evidence that a programme is funded, not that a capability exists. Both Blue Origin and ICON pass the first test and neither supplies the second.

Frontier Two variables that every architecture assumes and no source quantifies. The first is autonomy: every plan assumes lunar manufacturing runs uncrewed for long periods, and no source in this pack gives a mean time between interventions for any regolith-processing hardware, in vacuum, at temperature, in dust. Name that as an absence rather than filling it with an estimate. Established The best proxy for what will actually break is the Apollo dust record: all environmental and gas sample seals failed; suits were “more worn after 8 hr of surface activity” than training suits after 100 hours; zippers, wrist rings, faceplate mechanisms and cameras clogged; radiators and battery thermal control degraded. A sintering or excavation plant is a machine with seals, bearings, optical windows and radiators — every one of the Apollo failure modes. Mitigation maturity, from NASA Glenn: passive coatings and textured surfaces in ground testing, electrodynamic dust shields and lofting developmental, and “flight testing is the next step”. Frontier The second is optics in vacuum, which is specific to this slot. Azami and colleagues note that Rayleigh and Mie scattering effects for laser processing in the lunar vacuum remain “inadequately studied”, and solar sintering is limited by “low precision of its beam focus”. Both routes depend on delivering concentrated radiation through an environment full of charged, adhesive dust onto a surface that is being actively vaporised, and no source quantifies window or mirror lifetime.

Established The economic test for this slot is sharper than for the resource one, because the product is an object with a mass and a shipping cost. Delivered cost to the lunar surface is about $1.2 million per kilogram in 2023 pricing, audited by NASA's Inspector General. Imported material for one bagged, shielded module is about 0.7 tonnes — roughly $840 million of delivery charge, before the plant, the power or the crew. A 40 kWe reactor to run a high-temperature process is about ten tonnes — roughly $12 billion of delivery charge. Frontier Derived, and this is the displacement question stated as arithmetic: for local manufacturing to pay, the mass of plant plus power delivered must be less than the mass of finished goods it displaces, multiplied by the number of times it runs. A 10-tonne power plant plus, say, a 5-tonne fabrication unit is 15 tonnes delivered, about $18 billion, and to break even against shipping finished parts it must produce more than 15 tonnes of otherwise-imported mass over its life. At 233 kWh/t and 40 kWe it can heat at most about 1,500 t/yr of regolith, which is plenty of throughput and says nothing about yield. The binding unknown is not energy or strength; it is what fraction of processed regolith becomes a part that would otherwise have been imported. No source in this pack supplies that fraction. The general market question — who buys, at what price — belongs to Space Resource Economies.

Established And the programmatic record, which was set for this slot specifically and is unambiguous. The 2005 ISRU Capability Roadmap named surface manufacturing with in-situ resources and surface construction as two of its seven capability elements, set pilot-scale lunar construction for 2017, and assigned metal and silicon extraction TRL 2–3. Neither the construction milestone nor any surface fabrication has occurred. The Azami review notes that the carbothermal demonstration reached “technical readiness level to six” — but that is oxygen extraction, which is FR-II-01's territory, not fabrication, and no forming route has a stated surface-relevant readiness level above laboratory work on simulant. Frontier The current placement is the single most direct piece of evidence against the framing, and it comes from the agency running the programme. NASA's 2024 Moon to Mars Architecture puts “production of goods and services derived from lunar resources” in Sustained Lunar Evolution — the third of four segments — and does not require it for Human Lunar Return or Foundational Exploration. NASA's own plan does not treat manufacturing on the Moon as the point of going back.

Speculative Position one is the framing: lunar manufacturing is the purpose of returning to the Moon. Held by the ISRU advocacy tradition and parts of the commercial lunar sector. Against it: NASA's own 2024 architecture places resource-derived goods in the third of four segments and requires none of it for the first two. speculative as a statement about the actual programme; frontier as a statement about what would make a sustained presence affordable — and the second version is a good argument that the first version obscures.

Frontier Position two: sintered or melted regolith is the structural material of a lunar base. Held by most of the additive-manufacturing literature and by ICON and the agency construction programmes. For it: 4.2–31.4 MPa from laser powder bed fusion, up to 216 MPa from large-scale laser processing, 12–120 MPa from microwave sintering, 538 MPa from casting — all on simulant. Against it: the 233 kWh/t heating cost, a review that calls laser additive manufacturing “highly inefficient in the Lunar setting”, and vacuum optical effects that are unstudied.

Frontier Position three is the most under-covered live position in the subject: regolith bagging, not manufacturing, is the right answer for the near term. Held by Bao and colleagues by their own ranking — first place at 3.80 out of 5, 99% in-situ material, about 0.7 tonnes imported per module, and no high-temperature step. The argument that makes it decisive is that shielding is a mass-per-area requirement that 2–3 MPa satisfies. Frontier It is not a universal answer — pressure vessels, mechanisms, spares, electronics and plume-resistant landing surfaces all need something else — but it means that the first and largest structural need on the Moon is met by a technique that is barely manufacturing at all.

Established Position four: power, not process, is the binding constraint. Held explicitly by Azami and colleagues — “a major impediment is the need for power” — and supported by arithmetic: 233 kWh to heat a tonne to 1000 °C, a 40 kWe reactor massing about ten tonnes against a six-tonne design goal, and a derived ceiling near 1,500 t/yr heated with zero losses. Frontier The arithmetic is established and the causal claim is frontier, because bagging sidesteps it entirely and the routes that need power are the ones making high-strength parts rather than shielding.

Speculative Position five: the simulant evidence base transfers to the Moon. Held implicitly by every laboratory result in the field. Against it: NASA's own guide says agglutinates have no terrestrial analogue, nanophase iron cannot be made at scale, vacuum space-weathering and electrostatic charging are unreproduced, and microwave sintering strength varies tenfold with simulant choice alone. The honest position is that simulant results establish process plausibility, not performance, and that is a weaker claim than the field's citation practice implies. Speculative Position six: solar cells can be manufactured from regolith. Held explicitly by Blue Origin. Molten regolith electrolysis is real — the independent assessment puts it at about 16% oxygen yield and notes it has been demonstrated by NASA and Blue Origin — and it does produce silicon. What is missing is any published purity, efficiency or yield from the claimant.

Frontier Position seven: landing pads and roads are the first real product. Held by ICON and by most surface architectures, and the motivation is well attested — plume ejecta damage is why every architecture wants pads. Large-scale laser processing gives 56–216 MPa at 5 mm/min with a 3 kW laser, which is a real capability. Against it: nobody has stated a pad area, a laser power budget on the surface, or a build time, and the award announcement that funds the work states neither a readiness level nor a demonstration date. Frontier Position eight: autonomy is the real showstopper. Stated as a requirement everywhere and quantified nowhere in this pack. The evidence for taking it seriously is the Apollo dust record — every seal, bearing, window and radiator failure mode a plant would have. The brief's job is to say the number is missing rather than guess it.

Frontier Position nine: lunar manufacturing rehearses Martian manufacturing. Held by NASA's Moon-to-Mars framing. Against it: the chemistries barely overlap — lunar work is high-temperature reduction of silicate rock, and Martian in-situ resource utilisation as demonstrated is solid-oxide electrolysis of atmospheric carbon dioxide. What transfers is autonomy, dust handling and operations, not process. Worth stating because the transfer is usually asserted without qualification. Handwave Position ten: self-replicating or bootstrapping lunar industry. A long speculative tradition with nothing behind it beyond concept studies. State the idea in a sentence and route to Space-Based Manufacturing and Space Resource Economies; the arithmetic that matters is the displacement ratio in section 2, and nobody has supplied the yield fraction it needs.

Established The first bottleneck is energy, and it has a number. 233 kWh to heat one tonne of regolith to 1000 °C, before losses — roughly a 100 m² photovoltaic array running for six hours of lunar daylight — against sintering routes that actually operate at 1100–1350 °C. Frontier Derived, at the reference power scale: a 40 kWe reactor produces 350 MWh a year, which at 233 kWh/t and zero losses caps the process at about 1,500 tonnes of regolith heated per year. The reactor itself masses about ten tonnes against a six-tonne design goal and costs roughly $12 billion to deliver at the audited rate.

Established The second is that the entire evidence base is a proxy of measurable weakness. No simulant reproduces agglutinates, nanophase iron at 4–33 nm, vacuum space-weathering or electrostatic charging; NASA's guide instructs that at least two very different simulants be used for any single technology; and the same process spans 12 to 120 MPa on simulant choice alone. Every strength figure in the field should be read with an order-of-magnitude band attached.

Frontier The third is autonomy, and it is the bottleneck with no published number anywhere. No mean time between interventions exists for any regolith-processing hardware in vacuum, at temperature, in dust. Established What the Apollo record supplies instead is the failure-mode list a plant inherits in full: clogged mechanisms, abraded surfaces, failed seals, degraded radiators, scratched optics — with mitigation technologies in ground test and flight testing named as the next step.

Frontier The fourth is optical, and it is specific to this slot. Laser and solar routes deliver concentrated radiation through a vacuum full of charged adhesive dust onto a surface that is being actively vaporised. Rayleigh and Mie scattering in that environment is “inadequately studied”; solar sintering is limited by beam focus precision; solar sintering's only published strength, 2.49 MPa, is an ambient-pressure figure with no vacuum counterpart reported; and window and mirror lifetime is quantified nowhere.

Frontier The fifth is excavation resistance, where the one real datum went badly. LISTER reached about three feet against a planned ten in “really hard rock formations”. Frontier And the sixth is the number the economics actually turns on and nobody has: the yield fraction. A plant that heats 1,500 tonnes a year has enormous throughput; what matters is what fraction of processed regolith becomes a part that would otherwise have been imported, and no source in this pack supplies it. Fifteen tonnes of delivered plant and power costs about $18 billion and must displace more than fifteen tonnes of imports over its life, which is a yield question rather than a strength or energy question.

Established The adjudication carries three edges. Lunar Industry supplies the feedstock and the price, and the boundary is a unit operation. FR-II-01 owns prospecting, excavation, beneficiation, extraction chemistry, power system design, dust toxicology, delivered cost and the law. This brief takes what happens after there is feedstock. The two numbers that cross the boundary and must match on both pages are the audited $1.2 million per kilogram delivered and the 40 kWe reactor's roughly ten-tonne mass, because every calculation in this brief is denominated in them.

Frontier Space-Based Manufacturing owns microgravity fabrication and in-space assembly, and the split is physical rather than territorial. Powder handling behaves differently under one-sixth gravity than in free fall; thermal rejection differs between a structure radiating into space and one conducting into regolith; and lunar feedstock is a mineral mixture rather than a specified alloy or filament. Everything about assembling habitat-scale structures on orbit is FR-I-24's, and so is any claim about self-replicating or bootstrapping fabrication.

Frontier Space Resource Economies owns markets, property regimes, valuation and whether space resources have customers. This brief owns only the mass arithmetic — how many kilograms of plant and power displace how many kilograms of imports — and routes the moment the question becomes demand or price formation. That routing matters here more than in most briefs, because the displacement calculation has an unfilled term and filling it requires knowing what would have been shipped.

Established What this brief does not depend on is more chemistry or stronger material. Casting already reaches 538 MPa on simulant; large-scale laser processing reaches 56–216 MPa; and the requirement that dominates near-term demand — shielding — is met at 2–3 MPa. Strength is not the constraint. Power, autonomy, proxy fidelity and yield are, and none of them is a materials problem.

Established The experiment that would change this subject most is also the smallest: fabricate one object from real lunar material. Nothing has ever been sintered, printed, cast or formed from returned Apollo, Luna or Chang’e material beyond microgram-scale laboratory work. A single article made from returned regolith would convert the entire literature from process plausibility into performance data, and it requires no mission at all — only sample allocation.

Established Second, a methodological requirement that is already the published instruction and is routinely ignored: run every result on at least two very different simulants. NASA's own guide says so explicitly, and the one case where the field has honoured it produced the most useful uncertainty statement in the subject — microwave sintering spanning 12 to 120 MPa on simulant choice alone. Reporting a single-simulant strength without a second is now a known defect rather than a limitation.

Frontier Third: characterise solar and laser processing in vacuum, optically. Solar sintering's published strength of 2.49 MPa is an ambient figure with no vacuum counterpart; Rayleigh and Mie scattering effects in the lunar environment are “inadequately studied”; and window and mirror lifetime in charged adhesive dust is unquantified. Every optical route in the field rests on an unmeasured transmission path.

Frontier Fourth: build a landing pad, at a stated area, with a stated power budget and a stated build time. This is the one product with an attested near-term need — plume ejecta is why every architecture wants pads — a funded developer, and a candidate process at 56–216 MPa. None of the three numbers has been published by anybody, and a demonstration that produced them would be more informative than another compressive-strength paper.

Frontier Fifth: run regolith-processing hardware unattended long enough to measure a mean time between interventions. It is the number every architecture assumes and nobody has. The Apollo dust record says what will break; nothing says how often. Established And sixth, the one with a named remedy already: fly the dust mitigation hardware. NASA Glenn's status is passive coatings in ground testing and electrodynamic shields developmental, with the explicit statement that flight testing is the next step both to verify performance and to tailor the ground systems that everything else in this brief is tested on.

Established Take the requirement first and the process second, because doing it the other way round is how this subject overbuilds. The dominant near-term structural need is shielding, and shielding is specified in tonnes per square metre: 10–11 t/m² of regolith for a settlement dose target, or 8 g/cm² to halve galactic cosmic radiation and 50 g/cm² to quarter it. That requirement is met by 2–3 MPa material at 99% in-situ ratio with about 0.7 tonnes of imported bags and airbags per module, which is the top-ranked technique in the comparative literature and involves no furnace.

Frontier Where strength genuinely is required, the routes and their operating points are known. Laser powder bed fusion at 1100–1300 °C with scan speeds down to 1 mm/s gives 4.2–31.4 MPa on regolith alone and 264 MPa with an aluminium composite. Large-scale laser processing with a 3 kW CO₂ laser at 5 mm/min gives 56–216 MPa. Casting gives the highest measured figure at 538 MPa. Microwave sintering penetrates 65 cm at about 23% less energy than laser. Solar sintering reaches up to 1350 °C and 2.49 MPa at ambient. Every one of those is a simulant number and several are single-simulant numbers.

Established The power requirement is the one that converts a process choice into a mission architecture. 233 kWh per tonne to 1000 °C before losses; a 40 kWe reference reactor at about ten tonnes against a six-tonne goal, 1,250 kg of it lithium-hydride and tungsten shielding, with 133.4 m² of radiator at the lunar pole. Choosing a high-temperature route is choosing to deliver a reactor, and delivering a reactor is a $12 billion line item at the audited rate.

Frontier The environmental requirements are the ones with no numbers. Seals, bearings, optical windows and radiators must survive a dust environment that destroyed all environmental and gas sample seals on Apollo and wore suits more in eight hours than a hundred training hours did. Excavation must work against a surface whose only in-situ mechanical datum is a drill reaching a third of its planned depth. And optics must transmit through vacuum over a vaporising surface with scattering effects the review literature calls inadequately studied. None of those has a specification, and this brief names that rather than inventing one.

Established The seam with Lunar Industry is a process boundary and both pages carry it in the same words. FR-II-01 owns resource to commodity — prospecting, excavation, beneficiation, extraction chemistry, power, dust, transport cost, the economics and the law. This brief owns commodity to object — forming, sintering, printing, casting, bagging and assembly, and whether making a thing locally beats shipping it. The energy figures appear on both pages for different purposes: there they price a commodity, here they price a fabrication step. The two pages should not both carry the carbothermal table in full, and this one cites it in a line.

Established Historical Space Colonization Concepts owns the 1970s — the Ames summer study's lunar-material supply chain, the mass driver, and the solar-power-satellite economics that killed the funding case. This brief restates none of it, and the seam sentence is narrow: the idea that lunar material would be processed rather than shipped is fifty years old, its 1970s form belongs to FR-X-08, and what is new is that there are now measured compressive strengths, ranked routes and an audited delivery cost. In the other direction, FR-X-08 should not carry the 2024 solidification rankings, the simulant Figure of Merit or the CLPS cost.

Frontier Space-Based Manufacturing owns microgravity fabrication and orbital assembly, and the physical differences are named in section 1 rather than gestured at. Space Resource Economies owns markets and valuation. Orbital Shipyards is the in-space assembly counterpart, and the one sentence worth carrying is that a lunar factory's plausible highest-value product is structure for use off the lunar surface rather than on it — which is a claim about a customer, and routes.

Frontier And O'Neill Cylinders is where this brief's central finding does its most surprising work. That brief establishes that shielding is a mass-per-unit-area requirement at 6–7 t/m² of polyethylene or 10–11 of regolith, and that structure is not the constraint because a kilometre-scale habitat needs only about 200 MPa. Put those two findings beside Bao's ranking and the same conclusion appears at both scales: the hard part is moving mass into position, not making it strong.

Established The most direct institutional evidence against the framing comes from the agency doing the going back. NASA's 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 none of it for Human Lunar Return or Foundational Exploration. A brief claiming manufacturing is the point of the programme has to explain why the programme's own architecture document does not require it until the third phase.

Established The roadmap record is the second exhibit and it is specific to this slot. The 2005 ISRU Capability Roadmap named surface manufacturing and surface construction as two of seven capability elements, set pilot-scale lunar construction for 2017, and assigned metal and silicon extraction TRL 2–3. Neither the milestone nor any surface fabrication happened, and no forming route has a stated surface-relevant readiness level above laboratory work on simulant twenty-one years later.

Frontier The funding record is real and the capability record is not, and the two should be reported separately. ICON holds a roughly $57.2 million NASA SBIR Phase III award for lunar surface construction, announced in November 2022, with prior NASA and Department of Defense funding behind it and no stated readiness level or demonstration date. Blue Origin completed a Critical Design Review in September 2025 and plans a 2026 demonstration in a simulated lunar environment, with no purity, efficiency, yield or temperature disclosed. Both facts are established: the programmes are funded. Neither establishes a capability, and this brief states the grading rule rather than applying it silently.

Established The most useful institutional document in the subject is a technical memorandum about dirt. NASA's Lunar Regolith Simulant User's Guide tells the whole field, in writing, that no simulant replicates lunar regolith, that agglutinates and nanophase iron cannot be reproduced, and that at least two very different simulants should be used for any single technology. Interest running against the finding — it is the agency that funds most of the simulant work, telling everyone that the work has an unstated uncertainty band — and it is largely unobserved in the citation practice of the literature it governs.

Frontier And a governance observation that follows from the bagging result. If the top-ranked technique for the dominant near-term requirement is a bag and a shovel, then the institutional question is not which fabrication technology to fund but whether the fabrication framing is producing the right funding portfolio at all. A field that ranks its own options and finds the simplest one first, and then keeps funding the complicated ones, is telling you something about how research programmes select problems.

Speculative The ethical content here is mostly about claims rather than about harms, which is unusual and worth being explicit about. The public record of this subject consists substantially of vendor announcements describing capabilities without performance figures: solar cells with no efficiency, silicon with no purity, reactors with no temperature, construction systems with no readiness level or date. Those announcements circulate as evidence that lunar manufacturing works. Frontier The remedy is the rule stated in section 2 — a vendor capability claim with no performance number in it is evidence that a programme is funded, not that a capability exists — and applying it evenhandedly to companies and agencies alike.

Established The one occupational hazard with a real record is dust, and it is Lunar Industry's to analyse. What belongs here is the manufacturing-specific version: a fabrication plant is a machine with seals, bearings, windows and radiators, operating in a material that scratched Apollo gauge dials unreadable and destroyed every environmental and gas sample seal flown. Maintenance of that plant means human exposure, under a permissible limit of 0.3 mg/m³ derived from a ground-milled proxy whose representativeness its own source calls unknown.

Frontier A distinct and under-discussed risk is that safety-critical structures would be qualified on simulant data. A pressure vessel or a habitat wall made from lunar material would be certified against strengths measured on terrestrial proxies that do not reproduce agglutinates, nanophase iron, vacuum weathering or electrostatic behaviour — and the same process spans a factor of ten in strength on proxy choice alone. The honest engineering position is that the current evidence base cannot qualify a life-critical lunar structure, and the bagging route's appeal is partly that a shielding berm is not life-critical in the same way a pressure wall is.

Speculative And a resource-allocation question the ranking makes visible. If the dominant near-term structural need is met by bags and a shovel, then research funding directed at high-temperature forming is buying capability for the later, harder items — mechanisms, spares, electronics — rather than for the base. That may be the right choice and it is rarely stated as a choice, and readers deserve the ranking that makes it visible.

Frontier The civilizational case for making things on the Moon is a cost argument and it should be stated as one. Delivered cost to the lunar surface is about $1.2 million per kilogram; every kilogram made locally is a kilogram not delivered; and a sustained presence that imports everything is a presence whose cost scales with its size. That is a good argument, and it is a different argument from “manufacturing is the point of going back”.

Frontier The finding that reshapes the case is that the first and largest requirement is met without manufacturing. Shielding is a mass-per-unit-area requirement; the top-ranked technique in the comparative literature is bagging at 2–3 MPa and 99% in-situ material; and about 0.7 tonnes of imported bags and airbags supplies a module. The genuinely transformative step is therefore not a furnace but a loader, and the manufacturing routes become necessary for the second tranche of needs — pressure vessels, mechanisms, spares, electronics, landing surfaces — where lunar feedstock is least suited and the requirements are hardest.

Established What has genuinely changed, and is worth saying plainly, is that the subject now has comparative numbers rather than enthusiasm. Four families of solidification technique scored against each other on a common scale. A published heating energy per tonne. Compressive strengths with ranges. A simulant fidelity framework with eight scored properties and an explicit instruction to use two proxies. An audited delivery cost to denominate all of it in. None of that existed a decade ago, and it is what makes the honest verdict possible rather than rhetorical.

Speculative The terminal statement is a displacement ratio with one term missing. Fifteen tonnes of delivered plant and power costs about $18 billion and must displace more than fifteen tonnes of imports over its life to pay. It can heat about 1,500 tonnes of regolith a year, which is ample throughput. What nobody has published is the fraction of processed regolith that becomes a part which would otherwise have been shipped — and that single unpublished fraction, not energy and not strength, decides whether lunar manufacturing is an industry or a demonstration.

These horizons track fabrication specifically — not extraction, which belongs to Lunar Industry — and the base rate is poor: pilot-scale lunar construction was scheduled for 2017:

• 10 yr: Frontier Expect the first structure built from lunar material to be a shielding berm or a bagged enclosure rather than a printed or sintered article, because it is the top-ranked technique, needs no high-temperature step and imports about 0.7 tonnes per module. Frontier Expect a landing-pad demonstration to be attempted, since it is the one product with an attested need, a funded developer and a candidate process — and expect the first published pad area, power budget and build time to be more informative than the strength figure. Speculative Expect no surface fabrication plant of any scale: no forming route has a stated surface-relevant readiness level above laboratory work on simulant, and the reference power unit costs about $12 billion to deliver.
• 25 yr: Speculative A working high-temperature process on the surface requires the reactor, and the reactor requires the delivered-cost curve to bend, which it has not: the audited price rose 20% over the life of the contract designed to lower it. Frontier The most likely genuine capability at this horizon is construction rather than manufacturing — berms, pads, roads, buried habitat shells — using regolith in bulk with minimal processing. Speculative Fabrication of mechanisms, spares and electronics from lunar feedstock is a different problem and nothing in the current literature addresses it; the strength results all concern bulk structural material.
• 50 yr: Speculative A lunar factory that displaces meaningful import mass is coherent at this horizon if the yield fraction turns out favourable and if power delivery becomes affordable, and both are currently unknown rather than unfavourable. Speculative The plausible highest-value product would be structure for use off the lunar surface rather than on it, which makes the subject dependent on a customer that does not exist. Route to Orbital Shipyards and Space Resource Economies.
• 100 / 250+ yr: Handwave Beyond useful forecasting, and the honest statement is about what the constraints are made of. Strength is solved on simulant at every level the requirements need. Energy is arithmetic and improves with power. Speculative What does not improve on its own is the proxy problem: until something is fabricated from real lunar material, the entire evidence base remains a set of terrestrial experiments with an order-of-magnitude uncertainty band, and that is a sample-allocation decision rather than a technological horizon.

• Depends on Three edges. Lunar Industry supplies both the feedstock and the denominator: the boundary is a unit operation — resource to commodity there, commodity to object here — and the two numbers that cross it are the audited $1.2 million per kilogram delivered to the lunar surface and the reference 40 kWe reactor at about ten tonnes, in which every calculation on this page is denominated. Space-Based Manufacturing because the fabrication physics differs from microgravity work in three specific ways — powder handling under one-sixth gravity rather than free fall, thermal rejection into regolith rather than by radiation alone, and a mineral-mixture feedstock rather than a specified alloy — and because everything about assembling structures on orbit, including any self-replication claim, is FR-I-24's. And Space Resource Economies because the displacement calculation this brief performs has an unfilled term: what fraction of processed regolith becomes a part that would otherwise have been imported, which is a question about demand as much as about yield.
• Requires (not on this map) The first needs no mission, only a sample allocation: nothing has ever been sintered, printed, cast or formed from Apollo, Luna or Chang'e material beyond microgram-scale laboratory work, so every compressive strength in the literature is a terrestrial proxy figure — and NASA's own guide says agglutinates have no terrestrial analogue, nanophase iron cannot be synthesised at scale, and the same process spans 12 to 120 MPa on simulant choice alone. The second is the number every architecture assumes and none publishes: no mean time between interventions exists for any regolith-processing hardware operating in vacuum, at temperature, in dust, against a failure-mode list that Apollo demonstrated in full.
• Enables No typed enabling edge is claimed, and the programmatic record is why. No object has ever been fabricated from lunar material anywhere, every published strength is a simulant figure, no forming route has a stated surface-relevant readiness level above laboratory work, pilot-scale lunar construction was scheduled for 2017 and did not happen, 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. A capability that has never operated does not get to claim what it will unlock.
• Adjacent O'Neill Cylinders, where the same finding appears at a different scale: shielding is a mass-per-unit-area requirement of 6–7 t/m² of polyethylene or 10–11 of regolith, and structure needs only about 200 MPa, so at both scales the hard part is moving mass into position rather than making it strong. Historical Space Colonization Concepts owns the 1970s form of the same idea. Orbital Shipyards is the plausible customer for a lunar factory's highest-value product. And Mars Colonization is where the claimed rehearsal fails: the two chemistries barely overlap, and what transfers is autonomy, dust handling and operations rather than process.

Frontier “3D printing habitats out of regolith is the obvious way to build a Moon base.” The only peer-reviewed comparative ranking of the options puts regolith bagging first, at 3.80 out of 5, ahead of casting at 538 MPa, solar melting, direct sintering and microwave sintering — on a composite of in-situ material ratio, time, temperature requirement and equipment. Bagging gives 2–3 MPa, uses 99% local material, needs no high-temperature step, and imports about 0.7 tonnes per module. Frontier The reason it wins is that the dominant near-term structural need is shielding, and shielding is a mass-per-unit-area requirement rather than a strength requirement — 10–11 tonnes per square metre of regolith for a settlement dose target, or 50 g/cm² to quarter galactic cosmic radiation. You do not need to manufacture anything to make a shielded lunar habitat. You need bags and a shovel.

Established “We've already made bricks and parts from lunar regolith.” Not from lunar regolith. No object has ever been fabricated from lunar material, on the Moon or on Earth, beyond microgram-scale laboratory experiments on returned samples. Every compressive strength in the literature — 4.2–31.4 MPa, 56–216 MPa, 12–120 MPa, 538 MPa — is a simulant figure.

Speculative “Simulant results tell us what will happen on the Moon.” NASA's own guide is unambiguous: agglutinates “have no terrestrial analog” and cannot be reproduced at scale; nanophase iron at 4–33 nm is “essentially impossible to synthetically create in quantities matching lunar samples”; surface patinas, ion implantation, atomic displacement and vacuum electrostatic charging are not reproduced; solar-wind hydrogen and helium are absent; and “no lunar simulant completely replicates all aspects of the lunar regolith”. Frontier The concrete demonstration is that microwave-sintered compressive strength spans 12 to 120 MPa depending on which simulant is used — a tenfold spread from the proxy, not the process. The guide instructs that “at least two very different lunar simulants should be used” even for a single technology, and the literature routinely reports one.

Frontier “Energy is not really the issue — there's plenty of sunlight.” Heating one tonne of regolith to 1000 °C takes at least 233 kWh before losses, roughly a 100 m² photovoltaic array running for six hours of lunar daylight — and the sintering routes operate at 1100–1350 °C. Frontier Derived: a 40 kWe reactor makes 350 MWh a year, capping the process at about 1,500 tonnes heated per year with zero losses; the same energy at the carbothermal demonstration's 13 g O₂/kWh gives about 4.6 tonnes of oxygen a year, against modelled surface demand cases of 8.5, 17 and 34. Established And the review that surveys the laser routes calls them “highly inefficient in the Lunar setting” on the power conversion alone.

Speculative “Blue Alchemist has demonstrated lunar solar cells.” The company's own release claims iron, silicon, glasses, ceramics, solar cells, wire, oxygen, metals and semiconductors, landings “up to 60% cheaper”, a 3-acre facility with 65+ experts and a September 2025 Critical Design Review. It states no silicon purity, no cell efficiency, no oxygen yield, no reactor temperature and no award value. Frontier The independent check says molten regolith electrolysis is real — about 16% oxygen yield, fourth of eight at a score of 42.30, “demonstrated by NASA/Blue Origin” — so the process exists and the solar-cell claim built on it is where the evidence thins.

Frontier “ICON is building the lunar construction system, so the capability exists.” ICON holds a roughly $57.2 million NASA SBIR Phase III award announced in November 2022 for landing pads, blast shields, roads and habitats, on top of earlier NASA and Department of Defense funding. The announcement states no technology readiness level and no target date for a lunar demonstration. Frontier The general rule, stated so a reader can apply it: a vendor capability claim with no performance number in it is evidence that a programme is funded, not that a capability exists.

Frontier “Solar sintering is the low-tech answer — free energy, no laser.” Its only published compressive strength is 2.49 MPa at ambient pressure, with no vacuum figure reported, it is limited by “low precision of its beam focus”, and the vacuum optical physics — Rayleigh and Mie scattering in the lunar environment — is described in the review literature as “inadequately studied”. It is roughly fifty times weaker than large-scale laser processing, and its performance in the environment it would actually operate in has not been measured.

Frontier “We understand how hard the ground is.” The only in-situ mechanical datum anyone has is LISTER on Blue Ghost Mission 1, which reached about three feet against a planned ten after hitting “really hard rock formations”. Every excavation force, bucket-fill factor and specific energy in the design literature is a simulant number, and the one real measurement went the wrong way.

Frontier “Lunar manufacturing rehearses Martian manufacturing.” The operations transfer — autonomy, dust handling, thermal design, cost discipline. The chemistry does not: lunar work is high-temperature reduction of silicate rock and Martian in-situ resource utilisation as demonstrated is solid-oxide electrolysis of atmospheric carbon dioxide. Handwave And “self-replicating lunar industry” has nothing behind it beyond concept studies; the arithmetic that would matter is the displacement ratio, and nobody has published the yield fraction it needs.

Established And the framing itself. “Manufacturing on the Moon is the point of going back” fails on two independent lines. Frontier The engineering line: the top-ranked technique in the published comparative literature is barely manufacturing at all, and it wins because it meets the actual requirement. The programmatic line: NASA's own 2024 architecture places resource-derived goods in Sustained Lunar Evolution, the third of four segments, and requires none of it for the first two. Established The honest version is a good claim and a narrower one: making things on the Moon is what would make a sustained presence affordable, the forming routes are real and now comparatively ranked, and the field has never once run any of them on lunar material in place. “The point of going back” is a stronger claim, and the agency doing the going back has not made it.