No cities on the Moon: a billion tons of water is not enough for sustainability

1 Introduction

Can the Moon’s resources support cities? The Moon, with an area of 37.93 million km², about 50% bigger than North America (24.71 million km²) is talked about as our “8^th continent” ¹ . As a result, many authors consider the Moon’s resources to be so vast as to pose no significant constraints on industry and settlement (Wingo, 2004). Jeff Bezos ² has promoted moving heavy industry to the Moon and Elon Musk ³ has called for “self-growing” cities on the Moon. Presumably, in order to make an economic difference to the human economy, they are both imagining significant settlement of, say, at least 1% of Earth’s population, i.e., ∼80 million people living on the Moon. These are grand plans. They must however be grounded in the realities of the Moon’s resources. In particular, their proposed enterprises will require significant supplies of power and water to sustain these lunar settlements. Is it correct to make the assumption that sufficient power and water are available?

This paper compares the power and water resources of the Moon with the needs of a city. Cities are usually understood to be sustainable long-term urban areas with a population of at least 100,000, while a population of one million people would undoubtedly qualify a settlement as a city. Both cases are considered here. The three key questions are: “How much power?” “How much water?” and “How long qualifies as sustainable?” These questions are investigated in this paper ⁴ .

The lunar poles are where almost all the water on the Moon is located. The regions within a few 100 km of the poles also include specially favored regions where solar photovoltaic power can be tapped at scale. These two polar peculiarities are connected.

These High Illumination Regions are sometimes called the “Peaks of Eternal Light” (Elvis et al., 2016), though this is not quite accurate. The “Peaks” are found near to the lunar poles because the Moon has an obliquity to the plane of the ecliptic (the plane of the Earth-Moon orbit around the Sun) of just 1.5° (Allen, 1973), rather than the 23.5° of the Earth. As a result, the Sun always circles close to the horizon near the lunar poles and high points are rarely shielded from direct sunlight except by some other high points (Noda et al., 2008; Gläser et al., 2024). The low angle of the Sun allows photovoltaics to be mounted in vertical arrays. These arrays can be tall, from several meters to as high as 1 km (Ruppert et al., 2021) and could rotate to follow the Sun around the horizon.

The sunlit crater rims of the Peaks cast long shadows across nearby craters due to the low elevation of the Sun near the poles. These shadows create “permanently shadowed regions” (PSRs) on crater floors that have not seen direct sunlight for some 4 billion years and so could be accurately described as pits of eternal darkness. These “Pits” typically lie a few kilometers below the Peaks. Many of the PSRs have evidence for water (from neutron ⁵ absorption measurements (Sanin et al., 2017). A “cold trap” is a PSR that never gets above 110 K ⁶ . Although over most of the Moon’s surface liquid water evaporates rapidly in the near-vacuum of the Moon, at these cryogenic temperatures water ice sublimates away at rates of no more than 1 mm per billion years (Zhang and Paige, 2009; Zhang and Paige, 2010). Below this temperature, then, any water ice that fell into a PSR from comet or asteroid collisions in the early years of the solar system would have accumulated there and still be present.

2 Power

Power use per person in the USA in 2020 was 12 kWh/day for residential use, ranging from 6–10 kWh/day (0.25–0.80 kW) per person ⁷ . The high usage rates are found in states where heating or cooling is most important: for example, North Dakota and Louisiana respectively. The nearly 300° temperature range on the Moon (95 K–387 K, Williams et al., 2017) will likely require more power per person. Conservatively taking the high-end value for the USA, a city of one million people would need 800 MW.

Power would also be needed for growing food. US agriculture used 800 trillion BTUs in 2012 for a population of 314 million ⁸ (Suits and Moyer, 2024), i.e. 84.3 MW per million people or a quite modest 64W per capita. On the Moon additional power would be needed to maintain a suitable temperature and appropriate lighting for crops, unless the farms are housed in a transparent surface building in a high illumination region, though these regions are a scarce resource (Elvis et al., 2016). To be conservative we assume no heating and lighting is required for agriculture.

Industrial usage of energy in the USA is ∼1.6 times greater than residential usage ⁹ . Adding in that usage rate would increase the lunar power needs for a million people to ∼2.1 GW.

Data centers in space, including on the Moon, are being discussed ¹⁰ . They each need GW-level power and so could quickly dominate power usage on the Moon.

These power usages are given in Table 1.

TABLE 1. Power usage.

omitting heat and light.

There are only two presently available power sources that are useable on the Moon: solar photovoltaics and nuclear fission reactors.

The maximum amount of solar power available on the lunar South pole has been investigated in some detail (A.K. Ross et al., 2023). With the Sun permanently close to the horizon, the answer depends on the height of the solar panel arrays. With the low gravity of the Moon heights of at least 1 km ¹¹ are plausible (S. Ruppert et al., 2021). At this height solar panels could provide a mean of about 3 GW for 90% of the time, though the power varies by a factor 2 depending on the Sun’s location when allowing for shadowing of the arrays by other arrays (See Tables 5, 7 in Ross, et al., 2023).

Fission powered nuclear reactors are being investigated as power sources on the Moon ¹² . Reactors have the advantage of not being limited to the high illumination regions near the lunar poles. They do, however, have a need for substantial heat dissipation. A reactor producing 40 kW of electricity (kWt) needs 128 m² of water-cooled radiators (Mason et al., 2010), slightly more than half the 200 m² needed for a solar installation with the same power output (at an assumed a high end 30% efficiency ¹³ given the 1.37 kW/m² of incident sunlight ¹⁴ (Ross, et al., 2023.) If this requirement scales with power, then 3 GW of fission power will require 7.5 km² of radiator. PSRs have a view of a dark sky lit only by starlight and so would make good locations for radiators. There are over 400 PSRs with a large enough area to contain such an array of radiators (Paige et al., 2010). A 1 GW (electric) reactor uses ∼24 tons of fuel per year ¹⁵ implying a sizeable mining or import operation.

If water can be transmitted long distances (100 s km), then large lava tubes are an alternative habitation site (Feng et al., 2024). The more moderate temperatures in lava tubes (Haruyama et al., 2012) may lessen power requirements.

If industry requires no more than double the personal use rate, then a million people can be supported on the Moon with either solar or fission power. Supporting multiple millions of people will require fission power. For both solar and fission sources power generation is a weak constraint on the lunar population.

3 Water

3.1 Total lunar water mass

Estimates of the mass of lunar water are based on three techniques that have different sensitivities: (1) ultraviolet, optical and near-infrared reflectivities have characteristic signatures of water from the top few microns to millimeters of the surface, a frost layer. (2) Neutron absorption probes depths of 1–2 m by measuring how many hydrogen atoms are needed to reduce the cosmic-ray-induced neutron flux, but have all had many kilometer resolution. (3) Water or hydroxyls are the most likely molecules to contain hydrogen. Lastly, ground penetrating radar can probe depths of 10–20 m and is most sensitive to blocks of ice rather than veins of ice distributed among rock.

The earliest solid claim of the detection of water on the Moon was by Spudis et al. (2013). They used the Mini-SAR (synthetic aperture radar) instrument on the ISRO Chandrayaan-1 lunar orbiter ¹⁶ to infer the presence of several hundred million tons of water in the PSRs around the lunar South Pole.

Many other papers followed. For example, Fisher et al. (2017), Nozetter et al. (2001), Hayne et al. (2015) used the Lyman Alpha Mapping Project (LAMP) instrument on NASA’s Lunar Reconnaissance Orbiter (LRO) to measure the ultraviolet reflectance of the permanently shadowed regions using starlight and an emission feature from interplanetary hydrogen that both faintly illuminate the PSRs. Li et al. (2018) used the near-infrared spectra from the Moon Mineralogy Mapper (M3) spectrometer on Chandrayaan-1 to detect water absorption bands. Not all the studies are strongly positive. Recently, Li et al. (2026) used the ShadowCam optical imager (Robinson et al., 2023) to confirm that water is present, but in only 3.5% of cold traps, though these may contain 30 wt% of ice, mixed with regolith.

The collimated LRO Lunar Exploration Neutron Detector (LEND) provided low resolution (FWHM 10 km) maps of neutron absorption Mitrofanov et al. (2010) but suffered from instrumental background problems that made the effective resolution much worse (Lawrence, 2017), possibly diluting the signal. A complex study of the LEND data finds that distributions limited to cold traps fit the data (Elphic et al., 2007).

In a meta-analysis of previous studies, Brown et al. (2022) studied 65 large PSRs (39 of them Southern). Those that were “cold traps” (T < 110 K), and so could be stores of water ice, were ranked high. Those with a median temperature <140 K, for which stable water ice could exist 2.5 m below the surface (Siegler et al., 2016), were ranked intermediate. Brown et al. estimated the probable volatile content of the cold traps from multiple data sources (their Table 2). Eight craters ¹⁷ likely contain the richest stores of volatiles, including water. The total water content of all eight is estimated to be of order 34 million tons (see their Table 9). P. Hayne et al. (2021) note that there are many “micro” cold traps where water could be sequestered near the lunar poles, potentially expanding the supply, although the total sub-110 K area of these micro cold traps is a modest ∼10% of that of the larger PSRs. The total of 34 million tons is a factor ∼20 less than the initial Spudis et al., estimate.

TABLE 2. Water usage.

It is worth noting that the fraction of water that can be recovered will surely not be 100%. Oil fields, for example, have recovered fractions of 33%–50% (Speight, 2016). Achievable retrieval fractions for lunar water will only be known once we have ground truth on how the water ice is mixed with the surface regolith.

The water mass estimates are quite uncertain. The Elphic et al. (2007) study reports a factor three range of water content for Shackleton crater from 1.6 to 4.5 million tons depending on assumptions. While the Brown et al. total of 34 million tons might be low by a similar factor, an order of magnitude larger total water mass would seem to be a stretch, though. More detailed searches, particularly with neutron absorption at higher spatial resolution are really needed.

One billion tons of water will be taken here as a generous baseline. All the results can be readily scaled from this baseline.

A billion tons of water sounds like a large amount. It is one cubic kilometer. To put that volume on a familiar scale, that is enough to fill the area of Central Park in Manhattan to a depth of 120 m ¹⁸ , 30% taller than the Statue of Liberty. It is enormously more water than was known on the Moon before 2013. Whether lunar water is in practice abundant depends on context. The relevant context is explored below.

3.2 Water usage

Water usage for a lunar city will come from several needs: personal usage for drinking and hygiene, growing food, rocket fuel and, unlike on Earth, breathing. All these human activities will necessarily take place in a built environment, given the lack of any substantial atmosphere on the Moon.

Although there is a wide range, the personal usage rate of a typical US person is roughly 125 tons of water per year (“80–100 gallons/day/person” ¹⁹ . i.e., 125 million tons/year for a population of 1 million.

A person breathes in 6 L of air per minute at 15 °C (Pleil et al., 2021) weighing 7.2 g (i.e., 62 kg/day), and absorbs 5% of that mass in O₂, i.e., 3 kg/day/person or 1.1 t/year/person. A water molecule weighs 3 x 10⁻²³ g and two H₂O molecules are needed to produce one O₂ molecule weighing 5 x 10⁻²³ g, this O₂ usage rate implies a water consumption of 1.8 t/year/person to replenish the O₂ in the lunar habitat. That is only 1.4% of the direct water usage. The carbon dioxide that people breathe out will be recycled by plant crops, at some efficiency. Other systems may be needed to maintain a healthy CO₂ content in the air.

A permanent settlement, unlike the ISS, would have to grow food, which requires additional water. A World Bank estimate ²⁰ finds that 2–5 tons (2,000–5,000 L) of water are needed to feed one person for 1 day. That comes to 730–1825 tons/year/person, or ∼6–15 times as much as direct personal water consumption. Assuming, as an example, that due to improved cultivation methods, food requires just half the low-end estimate, i.e., three times as much water as does personal consumption, then people would use 500 tons/person/year of lunar water in total.

A much-invoked use for lunar water is to refuel spaceships by splitting H₂O into hydrogen and oxygen, that can both then be liquified and used to fill up propellant tanks (Sowers and Dreyer, 2019). This application has appeal as it could be the basis for a commercial industry on the Moon.

Propellant production is not overly demanding on lunar water. SpaceX Starship second stage uses 1,200 tons of propellant ²¹ and, once used, that propellant is lost and cannot be recycled. Still 1 billion tons of water could fuel almost a million rockets, or ∼3 per Earth-day for 1,000 years. In this case, a billion tons is effectively abundant. Note that although Starship uses methane, CH₄, and oxygen as propellants this does not change the numbers significantly. Methane can be made from the hydrogen in water plus carbon, that may well be available in the organics expected in the PSRs [Paige +2010], based on their being abundant in the carbonaceous meteorites and asteroids Ryugu and Bennu where complex organics such as amino acids and sugars are found (Furukawa et al., 2025).

At the usage rates in Table 2, a population of one million would exhaust the known water supply in just 2.4 years. Clearly efficient water recycling will be needed to support a city on the Moon.

3.3 Sustainable water recycling

Given a recycling efficiency, e, the time to fully deplete the entire water supply can be readily found. These times then let us consider what efficiency is needed for a sustainable settlement.

Let the initial extracted mass of water be M_i and the initial time to use up 100% of that mass be T. The usage rate M_i/T = Nm/T, where N is the population and m is the usage rate per person.

Water is used at a constant rate.

d M d t = − M i T (1)

but is recycled at a constant rate.

d M d t = + e M i T (2)

then the overall change in available mass is.

d M d t = − 1 − e M i T (3)

which is a constant depletion rate, leading to a purely linear decrease of mass versus time, and so.

M t = M i − 1 − e M i t T (4)

which reaches zero at.

t = T 1 − e (5)

The time to reach zero then depends on the efficiency, e. Table 3 summarizes the lifetime of the water supply for a range of recycling efficiencies for a small and a large city (10⁵ and 10⁶ people, respectively.) What is a plausible value for e?

TABLE 3. Years to full exhaustion of a billion tons of lunar water ^a .

assuming a usage rate of 500 tons/person/year.

The International Space Station (ISS) has one of the highest efficiency water recycling systems, the Environmental Control and Life Support System (ECLSS). Until 2023 this efficiency was 94%. At that efficiency a billion tons of water is gone in 40 years for a population of 1 million.

ISS water recycling efficiency was improved in 2023 to an impressive e = 0.98 per cycle. This was achieved by adding a Brine Processor Assembly (BPA) to the Urine Processor Assembly (UPA) (Gaskill 2023). That may sound like a small change, but it is a major achievement, cutting the losses threefold.

A large lunar base would surely have improved sanitation over the International Space Station (ISS), including the laundry and showers not included on the ISS, and toilets that work well in the 1/6 gravity of the Moon. These improved facilities may degrade the recycling efficiency.

An ISS-like 2% loss rate means that it would take 100 years for a city of one million people to use up a billion tons of water. For such a large investment, a lifetime of about a century seems to fall short of the sustainable, long-term settlement beyond Earth that some advocate. A small city of 100,000 people will instead take 1,000 years use up all the water. This timescale seems long enough to justify being called sustainable.

Water seems to impose a limit of a few hundred thousand on the long-term lunar population.

4 Potential solutions

There are potential ways to evade this population limit. First note that, while settlements in lava tubes can mitigate power usage, their water would still come from the same polar locations. Some losses in transmission seem inevitable and so the lava tube option does not obviously change the case for water significantly.

• More efficient recycling can stretch the timescales. To reach a 1000-year lifetime for a city of 1 million requires an 10-fold improvement in recycling efficiency over the ISS, i.e., e > 99.75. That goal will be hard to reach and to maintain, even for a disciplined and motivated population.
• Reduce water usage. Vertical farming ²² and a no/low meat diet may cut the water usage rate substantially. One vertical farm uses “97% less water than traditional farming” for strawberry growing ²³ . This would be a major factor if it can be applied to crops more generally. A vegan diet could save water as the water required to produce chicken is a factor two larger than soybeans per kilogram, while beef requires at least six times as much water as soybeans (Gerbens-Leenes et al., 2013). Hoekstra and Mekonnen (2012) show that the “water footprint” of a US individual is 2.8 tons/year, whereas for the low/no meat diet of India that is reduced to 1.1 tons/year. Lab grown meat may be more efficient and remove the pressure to adopt a vegan diet (Woll and Böhm, 2018).
• Import water to the Moon from other bodies. At ISS-like 98% efficiency a one-million-person city would require landing 10 million tons per year. At 1,000 tons/lander (about 10 times the Starship Human Landing System capability ²⁴ that is 27 landers per Earth day. This is a substantial, but not inconceivable rate. The supply of easy to reach (low delta-v) asteroids may be a limiting factor (e.g., Elvis, 2014).
• Find more lunar water. As a billion tons of water is over 10 times the known water content of the lunar PSRs, this potential mitigation would seem unlikely to be realized. In the case that an unexpected reservoir of water is found, then doubling the total available water would double the longevity of lunar settlements. Existing estimates may well underestimate the total water supply (Cannon et al., 2020). The models that derive water content (“water equivalent hydrogen”, WEH) contain numerous parameters ²⁵ . These parameters may be updated, changing the WEH values up or, just as likely, down. While neutron absorption methods probe to depths of 1–2 m (Sanin et al., 2017; McClanahan et al., 2024) and radar-based estimates probe down to 10 s m (Spudis et al., 2013), there may be water at greater depths that cannot be detected as yet. The orbiting Apollo 17 Apollo Lunar Sounder Experiment (ALSE) reached depths of as much as 1 km (Phillips et al., 1973). The depth reached depends on the material. Highly conductive material reduces the depth reached. In situ boreholes would, of course, be definitive for the locations sampled.

5 Implications

The limits explored above used a generous estimate of the amount of lunar water. The analysis showed that optimal management of the lunar water resource will depend greatly on how we expect the lunar population to grow in the next decades. This result precludes large-scale (>100,000) human habitation of the Moon, baring one of the potential solutions becoming feasible. If the settlement remains a “Moon Village” ²⁶ at a population of about 1,000, comparable to the winter-over population of Antarctica ²⁷ , then we can be relatively carefree in our use of lunar water. If the population reaches 100,000, then we need to start being careful. Populations of a million seem to be unsupportable for greater than about one century unless, again, one or more of the potential solutions are applicable. Note that similar, and perhaps more stringent, considerations will apply to “generation ships” for interstellar travel (Marin et al., 2018).

Many other factors will need to be considered to make a functioning Moon Village. These factors are beyond the scope of this study. Here we only wish to point out that a sustainable Moon settlement will necessarily be limited in population. Our intent is to ground proposals for human settlement on the Moon in the realities of the available lunar resources.

Our comparison of the long-term requirements of a city for lunar water to its actual availability highlight the need for stewardship of the lunar water supply. Humanity will need to plan, govern and regulate the use of lunar water, rather than adopt a ‘first come first served’, free-for-all approach to water extraction and use. If the lunar population is to grow beyond a hundred thousand, water management will be critical and challenging, and we should evaluate the options well in advance. It is our hope that this paper will stimulate thoughtful engagement by all lunar exploration stakeholders with the issue of lunar sustainability in general and long-term water stewardship in particular.

Correction note

This article has been corrected with minor changes. These changes do not impact the scientific content of the article.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. ME used ChatGPT to aid in the literature search

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: M. Reza Emami, University of Toronto, Canada

Reviewed by: Ildiko Tulbure, 1 Decembrie 1918 University, Romania

Akinjide Oluwajobi, Obafemi Awolowo University, Nigeria

Footnotes

Foing (2003) “Our 8^th Continent, the Moon” https://global.jaxa.jp/article/special/lunar/foing_e.html

“Jeff Bezos says factories must move to the moon”, A. Clark, MSN, 2025: https://www.msn.com/en-us/weather/topstories/jeff-bezos-says-factories-must-move-to-the-moon/ar-AA1QK4lC

Scientific American, 9 February 2026: https://www.scientificamerican.com/article/elon-musk-says-spacex-to-prioritize-landing-on-the-moon-instead-of-mars-city/

There is a reddit thread on this topic: https://www.reddit.com/r/space/comments/a4c72/water_on_the_moon_so_how_big_a_colony_can_it/

Specifically epithermal neutrons with energies >0.5 eV up to ∼10 keV, that are most readily slowed down by interactions with hydrogen [W.V. Boynton et al., 2012].

Degrees Kelvin, a temperature scale where absolute zero is 0 K (or minus 273 C).

2000–7,000 kWh/year, a mean of 12 kWh/day: https://www.eia.gov/todayinenergy/detail.php?id=49036

Susan Hicks, U.S. Energy Information Admin., “Energy for Growing and Harvesting Crops is a Large Component of Farm Operating Costs”: https://www.eia.gov/todayinenergy/detail.php?id=18431. 1 BTU = 2.9307x10−7 MWh

Wikipedia Energy in the United States: https://en.wikipedia.org/wiki/Energy_in_the_United_States

Woolacott, 2025, “The plans to put data centres in orbit and on the Moon: https://www.bbc.com/news/articles/cjewvpkw7weo

i.e., 20% taller than the Burj Khalifa skyscraper in Dubai. At 829.8 m this is currently the world’s tallest structure.

NASA “Fission Surface Power”, https://www.nasa.gov/exploration-systems-development-mission-directorate/fission-surface-power/

“Photovoltaics”, Spectrolab. www.spectrolab.com/photovoltaics.html. “Space Solar Cells”, Azur Space, www.azurspace.com/index.php/en/products/pro ducts-space/space-solar-cells.

“Standard Solar Constant and Zero Air Mass Solar Spectral Irradiance Tables”, ASTM International. https://store.astm.org/e0490-00ar14.html

Wikipedia “Burnup”: https://en.wikipedia.org/wiki/Burnup

Chandrayaan-1: https://www.isro.gov.in/Chandrayaan-1_science.html

Amundsen, Cabeus, de Gerlache, Faustini, Haworth, Slater, Shoemaker, and Sverdrup.

The area of Central Park is 341 ha (843 acres). Wikipedia: https://en.wikipedia.org/wiki/Central_Park

USGS https://www.usgs.gov/water-science-school/science/water-qa-how-much-water-do-i-use-home-each-day#overview (scholarly articles searched give details but not a US average.) 1 US Gallon = 3.8 L.

Fujs and Kashiwase (2023): https://blogs.worldbank.org/en/opendata/strains-freshwater-resources-impact-food-production-water-consumption#:∼:text=While%20the%20UN%20calculates%20that,is%20also%20critical%20for%20electricity

Wikipedia “SpaceX Starship”: https://en.wikipedia.org/wiki/SpaceX_Starship

“Vertical Farming” (Wikipedia): https://en.wikipedia.org/wiki/Vertical_farming

Plenty Opens World’s First Farm to Grow Indoor, Vertically Farmed Berries at Scale in Chesterfield, Virginia https://chesterfieldbusinessnews.com/2024/plenty-opens-worlds-first-farm-to-grow-indoor-vertically-farmed-berries-at-scale-in-chesterfield-virginia/

Starship HLS (Wikipedia): https://en.wikipedia.org/wiki/Starship_HLS

Including: the influx spectrum of Galactic Cosmic Rays at 1 Astronomical Unit (AU), a uniform ferroanorthosite (FAN) regolith composition found in both polar regions, and a constant regolith temperature of 100 K for the upper meter [A.B. Sanin+ 2017].

Moon Village Association: https://moonvillageassociation.org/moon-village-association/

Wikipedia “Antarctica”: https://en.wikipedia.org/wiki/Antarctica