In-Situ Resource Utilization: The Rocks Are Ready, The Ice Is Not

In-situ resource utilization spent the 2010s as a slide. Across a decade of architecture studies, working groups, and roadmap revisions, ISRU was where space-program managers gestured to make the long-term math work. The mining diagrams were beautiful. The hardware was a 30-kilogram breadboard in a Hawaiian field test. That has changed.

The last eighteen months put ISRU hardware on the lunar surface for the first time, ran an oxygen-extraction reactor through a thermal-vacuum cycle that earned a Technology Readiness Level of 6, and produced an honest, signed assessment from NASA's own program lead that splits the field cleanly in two. Some of it works. Some of it does not yet have the data to work. The point of this note is to say which is which, with numbers, and to explain what that means for the architecture being built around it.

The headline read is straightforward. The rocks are ready. The ice is not. Oxygen and metals from dry regolith are now a manufacturing and integration problem, with multiple credible vendors and a near-term flight demonstration path. Polar water mining, the commodity that everyone wants because it is the propellant story, is still a prospecting problem. NASA's own May 2025 progress review puts it in flat language: water and volatile extraction is "lacking sufficient resource knowledge to proceed without significant risk."^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730} You cannot design a plant for ore whose grade you have not measured.

That distinction reorders the next decade of lunar logistics. It also explains why the first useful surface plant on the Moon is not going to be the one most of the popular press is selling.

The functional definition is unromantic. ISRU is any hardware or operation that takes locally available material at a destination and turns it into a commodity that the rest of the mission would otherwise have to ship up from Earth.^{2Sanders, G. & Kleinhenz, J. Moon to Mars In Situ Resource Utilization (ISRU) Status Update. Incheon, Korea, 4–6 November 2024. NTRS 20240013906. https://ntrs.nasa.gov/citations/20240013906} On the Moon, the candidate commodities are oxygen, water and hydrogen, bulk and refined regolith, raw metals (aluminum, iron, titanium, silicon), construction feedstock, and eventually plastics and biological precursors. On Mars, the same logic applies with carbon dioxide instead of oxide minerals as the dominant feedstock.

The reason any of this matters is propellant. NASA's own commodity prioritization holds that polar water provides 100 percent of chemical propulsion propellant mass once you have it (oxygen and hydrogen, electrolyzed and liquefied), while oxygen extracted from dry regolith provides 75 to 80 percent of propellant mass with the fuel still shipped from Earth.^{2Sanders, G. & Kleinhenz, J. Moon to Mars In Situ Resource Utilization (ISRU) Status Update. Incheon, Korea, 4–6 November 2024. NTRS 20240013906. https://ntrs.nasa.gov/citations/20240013906} Either commodity is enormous. Both substantially relax the launch-cost constraint that defines every architecture above low Earth orbit.

So the question is not whether ISRU matters. The question is which ISRU commodity reaches a useful production rate first, and what that does to the rest of the architecture.

Five lunar missions in 2025 carried ISRU-relevant payloads to the south polar region. Three failed in interesting ways. Two succeeded in technically smaller, strategically larger ways. None of them produced the headline story most of them were marketed against. All of them mattered.

Blue Ghost Mission 1, Honeybee Robotics PlanetVac, 2 March 2025. Firefly Aerospace's Blue Ghost lander touched down successfully and carried out a clean surface campaign. PlanetVac, a pneumatic regolith collector built by Honeybee Robotics, fired pressurized gas into the lunar surface and lifted material into an onboard container that was sieved and photographed in real time.^{3NASA. NASA Lander to Test Vacuum Cleaner on Moon for Sample Collection. https://www.nasa.gov/missions/artemis/clps/nasa-lander-to-test-vacuum-cleaner-on-moon-for-sample-collection/ — covering the Honeybee Robotics PlanetVac demonstration on Firefly Aerospace's Blue Ghost Mission 1, 2 March 2025.} This is the first time a deliberate regolith acquisition system has run on the Moon. No drilling. No scooping. A gas blast and a vortex. The technical claim is small. The strategic claim is large: the simplest possible ISRU front-end now has flight heritage.

IM-2 Athena and PRIME-1, 6 March 2025. Intuitive Machines' second CLPS lander carried NASA's Polar Resources Ice Mining Experiment. The ride included two ISRU instruments: TRIDENT, a one-meter auger drill from Honeybee Robotics, and MSolo, a mass spectrometer designed to identify volatiles released from drill cuttings. The lander touched down approximately 400 meters from its intended Mons Mouton site, ended up on its side in a small crater, and depleted batteries before the drilling sequence could complete.^{4NASA. NASA Receives Some Data Before Intuitive Machines Ends Lunar Mission. 7 March 2025. https://www.nasa.gov/news-release/nasa-receives-some-data-before-intuitive-machines-ends-lunar-mission/} Both instruments powered up and returned partial data. NASA's official line is that PRIME-1 met its technology-demonstration objective. The honest line is that the first US drill on the lunar south pole tipped over before it could drill, and we still do not have the volatile measurements the mission was built to make.

Resilience, ispace HAKUTO-R Mission 2, 5 June 2025. A separate commercial lunar lander carrying European, Japanese, and Taiwanese payloads. Hard landing during terminal descent. ispace's published technical analysis attributes the failure to a Laser Range Finder anomaly that delayed valid altitude readings, leaving the lander with insufficient deceleration time.^{5Spaceflight Now, ispace's Resilience lander crash lands on the Moon, 6 June 2025. https://spaceflightnow.com/2025/06/06/ispaces-resilience-lander-crash-lands-on-the-moon/ — and ispace, technical cause analysis released 24 June 2025.} The HAKUTO-R Mission 2 surface platform is gone from the 2025 and 2026 cadence.

Odin, AstroForge, 27 February 2025. A 100-kilogram asteroid prospector launched as a rideshare on the IM-2 vehicle. Targeted near-Earth metallic asteroid 2022 OB5 for a flyby in late 2025. Lost on 6 March 2025 due to ground-station and communication failures.^{6AstroForge mission overview and Wikipedia. https://en.wikipedia.org/wiki/AstroForge — Odin (formerly Brokkr-2) launched 27 February 2025 as an IM-2 rideshare; declared lost 6 March 2025.} AstroForge has committed to a follow-on, DeepSpace-2, against the same target.

The score: PlanetVac, fully successful technology demonstration. PRIME-1, partial. Resilience, lost. Odin, lost. The pattern is not coincidence. ISRU payloads are riding on the first commercial lunar landers, and those landers are still in the part of the learning curve where landing is the hard part. Hardware that works on the bench gets a fair chance to fail on descent. This will continue for two or three more cycles. None of these failures invalidates the underlying technology. All of them slow the calendar.

While the surface story was a mixed result, the laboratory results were not.

In August 2025, an integrated prototype consisting of a Sierra Space carbothermal reduction reactor, a NASA Glenn Research Center solar concentrator, precision mirrors from Composite Mirror Applications, and avionics and gas analysis from NASA Kennedy Space Center spent two weeks running in the thermal-vacuum chamber at NASA Johnson Space Center. Operating temperature ranged from -45 °C to 1,800 °C. Production exceeded program goals. The integrated stack was certified at TRL 6.^{7NASA Johnson Space Center. Sunlight Extracts Oxygen from Regolith Using Solar Chemistry. https://www.nasa.gov/centers-and-facilities/johnson/sunlight-extracts-oxygen-from-regolith-using-solar-chemistry/ 1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730}

The numbers worth holding on to:

• A single reactor melt cycle is equivalent to a continuous production rate of 140 kilograms of oxygen per year.
• Five consecutive melt cycles ran under simulated lunar vacuum and thermal conditions without operator intervention.
• Oxygen extraction efficiency was greater than 20 grams of O₂ per kilowatt-hour of thermal input.
• Oxygen yield was greater than 20 percent by mass of the regolith fed into the reactor.
• Carbon recovery (the carbothermal process consumes carbon to liberate oxide-bound oxygen, then must reclaim it) was greater than 99.7 percent.

A second technology, Molten Regolith Electrolysis, is being advanced in parallel by Lunar Resources and Blue Origin. The MRE proposition is simultaneous oxygen and metals (iron, silicon, aluminum) from a single high-temperature reactor. Lunar Resources processed 25 kilograms of lunar highland simulant over 36 hours of continuous operation (24 hours of electrolysis) at NASA Kennedy under lunar environmental conditions. Measured oxygen rate matched theoretical levels.^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730} Blue Origin's integrated MRE plus purification stack is scheduled for environmental testing in 2026.

A third technology, hydrogen and carbon-monoxide reduction of mare regolith (the original RESOLVE-era approach), sits at TRL 5 to 6 for equatorial sites. Pioneer Astronautics, now part of Redwire, demonstrated a system that combined size sorting and beneficiation, hydrogen reduction for oxygen, and a separate CO reduction step for metal extraction.^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730}

In other words, three independent oxygen-from-regolith technology paths have moved from breadboard to vacuum-chamber-qualified hardware in a single development cycle. Two of them produce useful metals as a co-product. None of them depends on water ice. All of them work on the dry highland regolith that covers the southern polar region whether the ice story pans out or not.

This is what "the rocks are ready" means in practice.

Now the other side of the split.

The lunar south pole sits in a permanent geometric trick. The Moon's axial tilt is roughly 1.5°, against Earth's 23.5°. Stand at the pole and the Sun never rises far above the horizon. A small set of high points (the Shackleton rim, parts of the Mons Mouton plateau) sees light for more than 90 percent of the year. A larger set of crater interiors sees light never. These permanently shadowed regions, "PSRs" in the literature, are some of the coldest places in the inner solar system. Cold enough that water and other volatiles deposited over geological time should still be there.

Should be. The honest state of the evidence is that we have orbital and impactor measurements (LCROSS, Lunar Reconnaissance Orbiter Diviner radiometer, the LRO neutron spectrometer, the ShadowCam instrument on Korea's Danuri orbiter) that strongly suggest water ice is present in PSRs in some form, at some concentration, in some distribution. We do not yet have direct, in-situ measurements of how much water there is per kilogram of regolith, what mineral or physical form it takes, how it is distributed laterally and with depth, or how it would behave under the heat of an extraction reactor.^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730} PRIME-1 was designed to answer some of these questions. PRIME-1 tipped over.

NASA's own assessment, restated for emphasis: the polar water and volatile commodity has "highly dependent" extraction performance on resource form, concentration, and distribution properties that are "currently unknown."^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730} You cannot economically design a mining plant against unknown ore.

That is not because nobody is trying. The ground-side technology stack for polar water has been advanced in parallel with the regolith stack. Two extraction approaches now sit at TRL 4 to 6: a contained subsurface heating concept (the Honeybee Planetary Volatile Extractor, in which a coring drill captures icy regolith inside its inner wall and then heats it from the inside while a cold trap captures sublimated volatiles) and a continuous regolith processing reactor (the Lunar Auger Dryer ISRU concept from NASA Johnson, cancelled in development before reaching TRL 5 due to budget cuts but technically validated). The PVEx hardware demonstrated extraction performance of 43 to 56 percent of contained water from cores 5 cm in inner diameter and 0.5 m long, with regolith assumed at 4 to 6 percent water by mass.^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730} On the capture and electrolysis side, OxEon's solid oxide electrolyzer has run for thousands of hours at TRL 5, each stack producing up to 0.9 kg/hr of oxygen and 0.12 kg/hr of hydrogen from electrolyzed water at lunar-relevant scale.^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730}

The bottleneck is not the plant. The bottleneck is the resource.

This is a consequential distinction because it changes who needs to build what next. A government with a flagship-class budget can afford to design a polar-water plant against a credible-but-uncertain reserve estimate, accept the risk of building the wrong reactor, and learn from the failure. A commercial operator targeting positive net present value cannot. Recent quantitative analysis of the value-of-information from lunar prospecting concludes that pre-exploration ISRU business cases require an internal rate of return on the order of 50 percent to absorb compounded resource and technology uncertainty. After a credible reserve assessment, that hurdle drops dramatically.^{8Coutts, S. & Sowers, G. Value of Information for Lunar Ice Exploration. New Space, 2025. https://journals.sagepub.com/doi/10.1089/space.2024.0045} In simpler terms, every dollar spent prospecting now is worth several dollars not spent over-engineering against unknown ground truth later.

Both pathways (regolith and ice) need a way to get the feedstock from the ground into the reactor. This is the part of the architecture that feels least like a research problem and is therefore most often skipped in business-case slides.

NASA's ISRU Pilot Excavator (IPEx) reached TRL 5 against a flight-relevant simulated mission profile: ten metric tonnes of loose regolith, moved across the lunar surface, with a 30-kilogram-class robotic excavator using counter-rotating bucket drums. The counter-rotation matters. It cancels reaction forces, which is what allows the machine to be 30 kg instead of 300 kg, which is what makes it feasible to land in the first place. The same problem at Earth gravity would call for an excavator at least an order of magnitude heavier.

For hard, icy regolith (the polar case), NASA ran the Break the Ice Lunar Challenge through 2024. Fifteen teams demonstrated excavation rates of 12,000 kilograms over 15 days (800 kg/day) with each delivery requiring a 500-meter traverse, in a lunar-analog test environment. Six teams competed head-to-head at Alabama A&M for timed performance. The technology stack has crossed TRL 5 for hard icy material as well as loose regolith.^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730}

Both excavation lines are now at the point where the next investment is a flight-relevant integrated test, not another bench iteration.

ISRU does not exist on its own. It only matters when a customer mission needs the commodity. In February 2026, the customer calendar moved.

NASA confirmed a substantial revision to the Artemis program. Artemis III, originally planned as the first crewed lunar landing of the modern era, was rescoped to a crewed test in Earth orbit. The mission will dock the Orion capsule with one or both of the SpaceX and Blue Origin lunar landers, exercise propulsion and life support, and qualify the new spacesuits.^{9NASA. NASA Strengthens Artemis: Adds Mission, Refines Overall Architecture. February 2026. https://www.nasa.gov/directorates/esdmd/nasa-strengthens-artemis-adds-mission-refines-overall-architecture/} The first crewed lunar landing is now Artemis IV, currently targeted for 2028. The Interim Cryogenic Propulsion Stage will be replaced for missions beyond the first three, and both the Exploration Upper Stage and Mobile Launcher 2 have been removed from the near-term plan.

This matters for ISRU in two non-obvious ways.

First, ISRU's first paying customer is now further away than its developers planned. The reduction in near-term lunar surface cadence reduces the urgency of having a propellant-grade ISRU plant operational by the end of the decade. The plant developers (Sierra Space, Lunar Resources, Blue Origin, Helios) still need to fly demonstrations against a real surface environment, but the pull from the human program is weaker than it was a year ago.

Second, and counter-intuitively, the rescope is a gift to the prospecting case. The two or three years between Artemis III and Artemis IV are exactly the window in which a serious resource assessment program could close the gap between "we believe there is water" and "we have measured the grade and distribution well enough to design a plant." NASA reactivated VIPER in September 2025 (the rover was cancelled in July 2024 after an estimated $505 million build cost and a forecasted$104 million further overrun, then revived as a CLPS task order to Blue Origin's Blue Moon MK1 lander, with a target landing of late 2027).^{10SpaceNews. NASA revives VIPER lunar rover mission with Blue Origin lander award. September 2025. https://spacenews.com/nasa-revives-viper-lunar-rover-mission-with-blue-origin-lander-award/} PRIME-1's volatile measurements, when they are eventually made by a lander that does not tip over, will be priceless. ESA's PROSPECT instrument is manifested for a CLPS landing in 2026. Korea's KIGAM LUVED, JAXA and ISRO's LUPEX, and China's Chang'e-7 mission with its hopping probe designed to enter a permanently shadowed crater are all in the next two years.^{2Sanders, G. & Kleinhenz, J. Moon to Mars In Situ Resource Utilization (ISRU) Status Update. Incheon, Korea, 4–6 November 2024. NTRS 20240013906. https://ntrs.nasa.gov/citations/20240013906 11The Planetary Society, Chang'e-7: China's water-hunting lunar south pole mission, https://www.planetary.org/space-missions/change-7 ; Space.com, Hopping robot will hunt for moon water on China's 2026 lunar mission, https://www.space.com/the-universe/moon/hopping-robot-will-hunt-for-moon-water-on-...}

If the prospecting campaign delivers, the polar water plant gets designed against measured ground truth instead of orbital inference. If it does not, the architecture defaults to oxygen and metals from highland regolith and ships hydrogen up from Earth. Either path is viable. Only one of them is currently de-risked.

The most under-reported development of the last eighteen months is the commercial side of ISRU is no longer entirely speculative. Real customers have signed real offtake-style agreements at non-trivial dollar values. None of these agreements proves that a lunar economy exists. All of them prove that prospecting and demonstration missions can now be financed against signed demand.

Interlune raised an $18 million Series A and a$5 million SAFE during 2025 and 2026, plus a $4.84 million grant from the Texas Space Commission to build a regolith simulant centre near NASA Johnson. The commercial side of the deal book is more interesting than the venture side. Bluefors, a Finnish manufacturer of dilution refrigerators (the workhorse hardware for superconducting quantum computing, where helium-3 is the current refrigerant of last resort), agreed to take delivery of up to 10,000 litres of helium-3 per year from Interlune between 2028 and 2037. The US Department of Energy contracted for three litres of lunar helium-3 by April 2029, for the DOE Isotope Program.^{12GeekWire. Interlune brings in fresh funding to support moon mining. 2026. https://www.geekwire.com/2026/interlune-funding-moon-mining-safe/ ; VC Tavern, Interlune Secures $18M to Advance Lunar Helium-3 Extraction and Wins Major Commercial Deals. https://vctavern.com/interlune-secures-18m-to-advan...} Interlune's roadmap calls for a multi-spectral mapper in 2025, a sample-gathering mission in 2027, and a pilot extraction plant in 2029.

Magna Petra signed a $22 million payload services agreement with ispace-EUROPE in October 2025 to fly NASA's MSolo mass spectrometer on ispace's Mission 3, as part of Magna Petra's HALO (Helium Availability of Lunar Origin) reconnaissance campaign.^{13Magna Petra Corp. Magna Petra Corp. and ispace-EUROPE Sign $22M Payload Service Agreement to Deliver NASA's MSOLO Instrument to the Lunar Surface. October 2025. https://magnapetra.com/magna-petra-corp-and-ispace-europe-sign-22m-payload-service-agreement-to-deliver-nasas-msolo-instrument-to-the-lu...} Two helium-3 prospectors, one mission, separately financed.

Helios (Israel) is the model worth studying most carefully. The company developed a molten regolith electrolysis reactor and discovered that the same architecture, applied to terrestrial iron ore, produces 99 percent pure iron at roughly 50 percent lower energy and 20 percent lower production cost than conventional steelmaking. Helios is now selling green steel on Earth while continuing to advance its lunar oxygen-and-metals system, with two flight demonstrations in development.^{2Sanders, G. & Kleinhenz, J. Moon to Mars In Situ Resource Utilization (ISRU) Status Update. Incheon, Korea, 4–6 November 2024. NTRS 20240013906. https://ntrs.nasa.gov/citations/20240013906} The dual-use thesis (the same reactor architecture serves a real terrestrial market and a long-bet lunar market) is the funding thesis. Capital is patient when patience is subsidized by a near-term revenue stream.

Sierra Space, Blue Origin, and Lunar Resources are the three principal US-domestic oxygen-from-regolith developers with plant-class ambitions; ICON is advancing lunar construction (Project Olympus, a laser-based regolith vitrification system) with roughly $60 million in NASA funding and a Blue Origin New Shepard suborbital flight in early 2025; Honeybee Robotics, now part of Blue Origin, has hardware on the lunar surface and follow-on missions manifested.

The combined thesis: ISRU is no longer a one-customer market with NASA as the only check-writer. There are signed offtake agreements, terrestrial co-products that justify capital intensity, and at least three independent integrated stacks moving toward flight.

Two parallel non-NASA threads should be on the watch list of anyone serious about this domain.

China is executing a parallel architecture on its own calendar. Chang'e-7, scheduled to launch in August 2026, will deliver an orbiter, lander, mini-hopping probe, and rover to the lunar south pole. The hopper is designed to descend into a permanently shadowed crater for direct volatile measurements; the primary objective of the mission is "direct evidence of water ice."^{11The Planetary Society, Chang'e-7: China's water-hunting lunar south pole mission, https://www.planetary.org/space-missions/change-7 ; Space.com, Hopping robot will hunt for moon water on China's 2026 lunar mission, https://www.space.com/the-universe/moon/hopping-robot-will-hunt-for-moon-water-on-...} Chang'e-8, in 2028, will test ISRU technologies including 3D-printed bricks from lunar regolith and is the technical predicate for the China-Russia International Lunar Research Station planned for the 2030s. If Chang'e-7 returns clean ice grade data and Chang'e-8 demonstrates regolith construction, the ILRS architecture moves from announcement to plausible.

Regulation has firmed up faster than most observers expected. The Artemis Accords, the US-led non-binding multilateral framework for space resource extraction, reached 64 signatories by April 2026, with seven countries (Finland, Bangladesh, Norway, Senegal, Hungary, Malaysia, the Philippines) added in 2025 alone.^{14NASA. Artemis Accords. https://www.nasa.gov/artemis-accords/ ; U.S. State Department. https://www.state.gov/bureau-of-oceans-and-international-environmental-and-scientific-affairs/artemis-accords} The Accords reaffirm that space resource extraction is consistent with the 1967 Outer Space Treaty's prohibition on national appropriation, draw the workable distinction between extracting a resource (allowed) and claiming sovereignty over the location (not allowed), and provide a basis for national legislation on private resource rights. The United States, Luxembourg, the United Arab Emirates, and Japan have such legislation in place. The legal floor is set; the regulatory ceiling, especially around tailings management, safety zones, and conflicting claims at high-value PSR locations, is not.

A short coda on the Mars side, because MOXIE produced the cleanest single ISRU result of the program era and its lessons travel.

MOXIE retired on 7 August 2023 after sixteen runs. Total oxygen produced: 122 grams, roughly the volume a small dog breathes in ten hours. Peak production: 12 grams per hour, twice the design target. Purity: 98 percent or better.^{15NASA. NASA's Oxygen-Generating Experiment MOXIE Completes Mars Mission. 7 September 2023. https://www.nasa.gov/missions/mars-2020-perseverance/perseverance-rover/nasas-oxygen-generating-experiment-moxie-completes-mars-mission/ 16Hoffman, J. A., Hecht, M. H., et al. Mars Oxygen ISRU Experiment (MOXIE) — Preparing for human Mars exploration. Science Advances, 2022. https://www.science.org/doi/10.1126/sciadv.abp8636} The point of MOXIE was never the volume. The point was the proof that solid-oxide electrolysis of CO₂, using Martian atmospheric pressure and Martian temperature swings as the operating envelope, behaves the way the bench models predicted. It does. The next step is a unit roughly 100 times larger that runs for a year on the surface and produces enough oxidizer for a Mars ascent vehicle. That is the architecture SpaceX has publicly committed to for crewed Starship return from Mars; whether it is built, by whom, and on what timeline is a separate set of questions, but the technical foundation MOXIE laid is no longer in dispute.

The pattern is the same on both worlds: the chemistry works, the prospecting and integration work do not yet have the answers they need, and the calendar belongs to whoever invests in resource assessment between now and the end of the decade.

The architectural conclusion of all of the above is more concrete than the field usually gets, so it is worth saying directly.

The first useful ISRU plant on the Moon should produce oxygen and structural metals from highland regolith, sized to support life support, EVA recharge, and surface mobility for a sustained crewed presence at a single site. Not propellant. Propellant requires hydrogen, hydrogen requires water, and water requires resource assessment we do not yet have. Oxygen-and-metals plants can be designed against a ground truth that is essentially understood (highland regolith composition has been characterized by orbital spectroscopy, returned samples, and the Chang'e-5 lab studies; carbothermal and MRE chemistry have crossed TRL 6).^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730 17Extracting Oxygen from Chang'e-5 Lunar Regolith Simulants. ACS Sustainable Chemistry & Engineering. https://pubs.acs.org/doi/10.1021/acssuschemeng.2c03545} They use roughly the same excavation and material handling stack as a future water plant, so the surface infrastructure investment is not stranded.

The polar water campaign should be a prospecting campaign first and a plant program second. VIPER, PRIME-1 follow-ons, ESA PROSPECT, JAXA-ISRO LUPEX, Korea KIGAM LUVED, China Chang'e-7. As many independent measurements of as many sites as the budget allows, before any commercial operator commits capital to an extraction plant. The economic argument here is not "wait." It is "the value of information from prospecting compounds. The value of information from a wrongly-sized plant is negative."^{8Coutts, S. & Sowers, G. Value of Information for Lunar Ice Exploration. New Space, 2025. https://journals.sagepub.com/doi/10.1089/space.2024.0045}

Commercial actors should target the dual-use thesis. Helios is the working example: a reactor architecture that earns terrestrial revenue while accumulating flight heritage. Interlune and Magna Petra are testing a different version of the same idea (terrestrial customers (quantum-computing refrigerant suppliers, the DOE) signing offtake agreements that finance lunar prospecting). The binary "moon mining either works or it doesn't" framing is wrong. The serious plays are companies whose unit economics close at terrestrial scale even if the lunar program slips by a decade.

Government programs should not over-specify the plant. Three independent oxygen-from-regolith chemistries are now at flight-demonstration readiness. The Lunar Infrastructure Foundational Technology missions (LIFT-1 and LIFT-2 in NASA's roadmap^{1Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730. https://ntrs.nasa.gov/citations/20250003730}) should fly competitive demonstrations, not a downselected single architecture. The cost difference between two parallel demonstrations and a single demonstration that turns out to be wrong is a factor of approximately twenty. The history of large-program premature downselection is not encouraging.

ISRU has crossed the line where engineering questions stop being theoretical. We know how to extract oxygen from highland regolith at TRL 6 with a production rate equivalent of 140 kilograms per year per reactor, and we know that the same machine can also produce iron, silicon, and aluminum. We do not yet know whether there is enough recoverable water in any specific permanently shadowed region of the lunar south pole to justify a propellant plant. The two facts, together, point at a near-term lunar surface architecture that runs on imported propellant, locally produced oxygen and metals, a sustained prospecting program targeting the polar volatile question, and a commercial sector whose business cases close on Earth before they need to close on the Moon.

That is not the lunar economy that gets pitched in keynotes. It is the lunar economy that gets built. The first useful plant on the Moon will be a logistics asset before it is a mining business, and the credibility of the entire field over the next five years will be determined by whether its proponents can hold to that distinction when the slide-deck version of the story is far more attractive.

The rocks are ready. The ice is not. Build accordingly.

---

Last revised 2026-05-04. Citations below trace every load-bearing claim to a primary or peer-reviewed source. Working materials, including the full NASA progress reviews, the bibliography, and notes, live in the post's working folder and are excluded from the public site.

1. Sanders, G. & Kleinhenz, J. Progress Review of NASA Lunar ISRU Development: 2019 to 2025. Luxembourg Space Resources Week, 19 May 2025. NASA Technical Reports Server, NTRS 20250003730.

2. Sanders, G. & Kleinhenz, J. Moon to Mars In Situ Resource Utilization (ISRU) Status Update. Incheon, Korea, 4–6 November 2024. NTRS 20240013906.

3. NASA. NASA Lander to Test Vacuum Cleaner on Moon for Sample Collection. — covering the Honeybee Robotics PlanetVac demonstration on Firefly Aerospace's Blue Ghost Mission 1, 2 March 2025.

4. NASA. NASA Receives Some Data Before Intuitive Machines Ends Lunar Mission. 7 March 2025.

5. Spaceflight Now, ispace's Resilience lander crash lands on the Moon, 6 June 2025. — and ispace, technical cause analysis released 24 June 2025.

6. AstroForge mission overview and Wikipedia. — Odin (formerly Brokkr-2) launched 27 February 2025 as an IM-2 rideshare; declared lost 6 March 2025.

7. NASA Johnson Space Center. Sunlight Extracts Oxygen from Regolith Using Solar Chemistry.

8. Coutts, S. & Sowers, G. Value of Information for Lunar Ice Exploration. New Space, 2025.

9. NASA. NASA Strengthens Artemis: Adds Mission, Refines Overall Architecture. February 2026.

10. SpaceNews. NASA revives VIPER lunar rover mission with Blue Origin lander award. September 2025.

11. The Planetary Society, Chang'e-7: China's water-hunting lunar south pole mission, ; Space.com, Hopping robot will hunt for moon water on China's 2026 lunar mission,

12. GeekWire. Interlune brings in fresh funding to support moon mining. 2026. ; VC Tavern, Interlune Secures $18M to Advance Lunar Helium-3 Extraction and Wins Major Commercial Deals.

13. Magna Petra Corp. Magna Petra Corp. and ispace-EUROPE Sign $22M Payload Service Agreement to Deliver NASA's MSOLO Instrument to the Lunar Surface. October 2025.

14. NASA. Artemis Accords. ; U.S. State Department.

15. NASA. NASA's Oxygen-Generating Experiment MOXIE Completes Mars Mission. 7 September 2023.

16. Hoffman, J. A., Hecht, M. H., et al. Mars Oxygen ISRU Experiment (MOXIE) — Preparing for human Mars exploration. Science Advances, 2022.

17. Extracting Oxygen from Chang'e-5 Lunar Regolith Simulants. ACS Sustainable Chemistry & Engineering.