Helium-3 and the Fusion Argument — What Is Strong, What Is Magical Thinking

Document 4 of the Moon set. The single technical claim used to justify the largest investments in lunar return: that lunar helium-3 will fuel fusion reactors on Earth. The strongest case (Kulcinski, Schmitt, the Wisconsin group); the strongest critique (Frank Close's "moonshine" line, the D-D side-reaction problem, the TU Delft 2014 economic study, the timing argument). The near-term quantum-computing market, where money is actually being committed in 2025-2026.

This piece is about a single technical claim, made in many places, used to support large investments and ambitious programmes. The claim is that lunar helium-3 will fuel fusion reactors on Earth. Strong versions of the claim say this will solve climate change, end the global energy problem, and pay for lunar industrialisation many times over. Weaker versions say it is a long-term option worth investing in now.

The publication's reading of the open literature is that the strong version is much weaker than its advocates present, and that the weaker version still requires more support than is usually offered. Both readings are presented at strength. The publication does not adjudicate the fusion physics; it lays out which physicists and engineers say what, and what the consensus has converged on so far.

The piece has three parts: what helium-3 is and where it is; the strongest case for lunar helium-3 mining; and the strongest critique. A closing section addresses the near-term commercial market — quantum computing — separately, because that market is real and is what is actually being transacted in 2025-2026.

What helium-3 is, and where it is

Helium-3 is the lighter, stable isotope of helium — two protons and one neutron, rather than the more common helium-4's two protons and two neutrons. It is one of only two stable nuclides with more protons than neutrons (the other being ordinary hydrogen). It is stable, non-radioactive, and behaves like an entirely different material from helium-4 in fundamental respects of quantum mechanics. It is, by any sensible measure, rare.

On Earth. Helium-3 occurs naturally at very low concentration. It is produced as a decay product of tritium; most current commercial supply comes from the decay of tritium stockpiles built up from Cold War-era nuclear weapons production. The US natural gas stream contains helium-3 at concentrations of 70-242 parts per billion of the helium fraction; the helium fraction is itself up to 7% of some natural gas wells. The total US stockpile of helium-3 from natural gas is in the range of 12-43 kilograms; annual separation from natural gas yields approximately 5 kilograms. Total global commercial supply is in the low thousands of litres per year, sold for high-value applications.

On the Moon. Helium-3 is implanted in the lunar regolith by billions of years of solar wind. The Moon has no magnetic field to deflect the solar wind, and no atmosphere to absorb it, so the surface layer of regolith contains an accumulated record of solar particles. Apollo samples and subsequent measurements show helium-3 at concentrations between 1.4 and 15 parts per billion in sunlit regolith. Permanently shadowed regions may contain higher concentrations — possibly up to 50 ppb — but this has not been directly measured. The total quantity in the top metre of lunar regolith is estimated at approximately 1 million tonnes, distributed across the lunar surface.

For comparison: the cited "1 million tonnes" is across approximately 38 million square kilometres of lunar surface. To extract it at the cited concentrations would require processing the top metre of regolith across most of the Moon. The accessible quantity at a given mining site is much smaller. Apollo samples at a single site cluster around 4-7 ppb on average. The widely-cited Chinese estimate that "three space shuttle missions per year could bring enough fuel for all human beings across the world" depends on extracting the helium-3 at concentrations far higher than have been observed and on a fusion technology that does not exist.

Elsewhere in the solar system. Jupiter's atmosphere contains helium-3 at concentrations of approximately 1 part in 10,000 of helium-4, which is itself a major atmospheric component. The total quantity in Jupiter's atmosphere is many orders of magnitude greater than on the Moon. Extraction is energetically prohibitive at current capabilities, but the long-run case for helium-3 mining tends to point at the gas giants rather than the Moon if the question is "where is the material in serious quantity?"

The strongest case for lunar helium-3

Three distinct arguments support lunar helium-3 mining. Each has different strength and depends on different assumptions.

Argument one: fusion. A fusion reaction between deuterium (the heavy isotope of hydrogen, abundant in seawater) and helium-3 produces a proton and a helium-4 nucleus, both charged particles. This is in contrast to the deuterium-tritium reaction (the basis of all current mainstream fusion research) which produces a neutron and a helium-4 nucleus. Neutrons are uncharged and therefore difficult to capture as useful energy; they also produce induced radioactivity in reactor walls. Charged particles can be captured directly as electrical energy through magnetohydrodynamic methods and do not induce wall radioactivity. The case for D-³He fusion is that it would be cleaner, simpler in some respects, and more efficient in energy conversion.

The advocates: Gerald Kulcinski at the University of Wisconsin-Madison, who has been the leading academic proponent for forty years; Harrison Schmitt, the Apollo 17 astronaut and geologist; the wider US "Fusion Technology Institute" community; and a number of Chinese and Russian researchers. Kulcinski's group built a small inertial-electrostatic-confinement reactor at Wisconsin and demonstrated helium-3 fusion at the laboratory scale, though without net power output. Schmitt has argued in successive papers and books that the case for returning to the Moon is, in the long run, the case for the energy resource.

The strength of the argument: the underlying physics is real. D-³He fusion is a known reaction. It does produce charged-particle products. It does avoid the neutron problem of D-T fusion. The fundamental physics is not the question.

Argument two: quantum computing and near-term commercial uses. Helium-3 is required for dilution refrigeration — the cooling technology used to reach the millikelvin temperatures at which superconducting and topological qubits stabilise. The technology has been used in research laboratories for decades; the commercial scale-up of quantum computing in the 2020s has created a substantial new demand. Quantum-computing companies are buying helium-3 in volumes that are now competitive with traditional research-laboratory demand. Helium-3 is also useful as a neutron-detection medium (for radiation monitoring, port security, scientific instrumentation), as a hyperpolarised contrast agent in MRI lung imaging, and in cryogenic neutron polarisation experiments.

The September 2025 agreement between Bluefors (a Helsinki-based cryogenics company supplying the quantum-computing market) and Interlune (a US-based lunar resources startup) is for up to 1,000 litres of lunar helium-3 per year, at a total contract value of approximately $300 million. The US Department of Energy in mid-2025 announced the first government procurement of an extraterrestrial resource: 3 litres of lunar helium-3, intended to seed an early supply chain. The commercial market is real. The numbers are small (1,000 litres per year is approximately a kilogram per year) but they are non-zero and they are paid for.

The strength of the argument: the market exists, the contracts are real, and demand from the quantum-computing sector is growing. If Earth-side supply from tritium decay cannot scale (and it cannot, because the tritium stockpile is finite and not being replenished at scale), lunar supply may be the only realistic option for the small but high-value market.

Argument three: option value and the spillover bet. Even if neither the fusion case nor the quantum-computing case is independently decisive, the case for lunar helium-3 may be made as an option-value play. Establishing extraction infrastructure now creates the capability to supply whatever future demand emerges; if fusion eventually works, the infrastructure is in place; if quantum computing scales, the infrastructure is in place; if other applications emerge, the infrastructure is in place. The technology spillover from developing extraction equipment — surface mining at vacuum, regolith handling, cryogenic processing — has terrestrial and Mars-mission applications.

The strength of the argument: option value is a real concept. The cost of having the capability and not needing it is lower than the cost of needing the capability and not having it. The same argument was made for some Apollo-era investments that produced unexpected spillover benefits.

The strongest critique

Three critiques cut against the fusion case in particular. Each is, in the publication's reading of the literature, harder to answer than its advocates often acknowledge.

Critique one: D-³He fusion is a self-defeating market.

This is the critique that, in the publication's view, lands hardest. It is straightforward and rests on physics rather than on engineering speculation.

D-³He fusion requires plasma conditions (temperature, density, confinement time) that are an order of magnitude more demanding than D-T fusion. Any reactor capable of D-³He fusion will, at the same conditions, support D-D fusion reactions as a side process. D-D fusion produces, among other products, tritium and helium-3. The reactor that successfully fuses deuterium with helium-3 will be producing helium-3 inside itself as a side product of the D-D reactions inevitable at those conditions.

A D-³He fusion plant capable of net power output will, on this analysis, be a helium-3 production facility through its own D-D side reactions. The reactor breeds its fuel. Establishing a lunar mining-and-transport infrastructure to supply helium-3 to a reactor that produces its own helium-3 is, in the published critique (New Space Economy, 2025; Dwayne Day, Space Review, 2015), "magical thinking."

The advocates' response: the breeding ratio may not be sufficient for self-sufficient fuel supply, and external supply may improve the economics. The critique's response: at the precision of current modelling, this is asserted rather than demonstrated, and the burden of proof should sit with the lunar-mining case rather than against it.

The publication's view: this is the strongest version of the critique, and the publication has not seen a response to it in the open literature that resolves it. The lunar helium-3 fusion case may turn out to be defensible, but on current public evidence, it depends on a fuel-breeding shortfall in D-³He reactors that has not been established. This is decision-relevant.

Critique two: the economic case does not close at credible scale.

The 2014 TU Delft study (Kulcinski-tradition friendly methodology, published in Acta Astronautica) modelled the economics of lunar helium-3 mining at various scales.

To supply 10% of global energy demand in 2040: 200 tonnes of helium-3 per year required. Regolith mining rate of 630 tonnes per second. 1,700-2,000 mining vehicles. Heating power of 39 GW. Power-system mass of 60,000-200,000 tonnes for the lunar operations. A fleet of three lunar ascent/descent vehicles and 22 continuous-thrust vehicles for orbit transfer.

At 1% of global energy demand: annual costs of €45.6-140.3 billion. Expected annual profits of -€78.0 to +€23.1 billion. The mission produces a net profit only in the best-case scenario, only at medium-to-large scale, and only with very large initial investment.

At 0.1% of global energy demand: annual costs of €7.7-20.5 billion. Annual expected profits of -€14.3 to -€0.8 billion. No profit even in the best case.

The TU Delft study's authors are not hostile to lunar helium-3 mining. The methodology is more favourable to the lunar case than most independent assessments. The conclusion — net profit only at scale, only in best-case assumptions, only with very large initial investment — is therefore the floor of the case, not the ceiling.

The advocates' response: the scenario is forty years out; the cost assumptions are wrong; private-sector launch costs have fallen by an order of magnitude since the study was published. The critique's response: the cost assumptions are not the binding constraint; the binding constraint is the energy balance — the energy required to mine and transport the helium-3 is, on the published modelling, a substantial fraction of the energy that the fusion reactor would produce from it.

The 1994 Wittenberg study found that extracting helium-3 from terrestrial interplanetary-dust sources consumed more energy than the fusion would release. The lunar case is more favourable than the terrestrial-dust case but not by margins that close the economic question at the published assumptions.

Critique three: the timing argument.

The case for lunar helium-3 mining as an "energy solution" depends on fusion working at commercial scale, on lunar mining infrastructure being deployed, and on the integration of the two — all by some date close enough to matter for present decision-making.

Mainstream fusion researchers' best guesses for commercial D-T fusion at meaningful scale: optimistic, 2040s; mainstream, 2050s-2060s; pessimistic, the second half of the 21st century. D-³He fusion is, on the same researchers' assessments, at least a decade further out than D-T, because the plasma conditions are an order of magnitude more demanding.

Lunar helium-3 mining infrastructure: optimistic, 2030s for demonstration; mainstream, 2040s-2050s for commercial scale; with the caveats from critique two about the economic case.

For the fusion case to be the case for lunar helium-3 mining, the two timelines need to align: fusion infrastructure ready to consume helium-3 at scale, lunar infrastructure ready to deliver at scale, both supporting the energy-mix-fraction case the advocates make. The earliest plausible alignment is the 2070s.

Present decision-making on lunar resource extraction is making a bet on technology and economics that are at least 50 years out from present capability. The advocates' response: this is the right time-horizon for civilisation-scale infrastructure investment. The critique's response: it is the right time-horizon if the assumption is that the technology will arrive; if the assumption is that the technology might not arrive, the investment looks different.

What Frank Close said

Frank Close is the Emeritus Professor of Physics at the University of Oxford, a former head of the Theoretical Physics Division at the Rutherford Appleton Laboratory, and the author of Lucifer's Legacy: The Meaning of Asymmetry, among other widely-read books on physics. In 2007 he wrote an article in Physics World titled "Fears over factoids" that addressed, among other things, the public framing of lunar helium-3 fusion.

Close described the lunar helium-3 fusion proposition as "moonshine." His argument:

1. D-³He fusion is harder than D-T fusion at the level of fundamental plasma physics, not easier.
2. The D-D side reaction inherent in any D-³He reactor produces tritium and additional helium-3 inside the reactor.
3. The energy balance of lunar mining-and-transport-and-fusion has not been demonstrated to be net-positive at the modelling precision available.
4. The public framing of lunar helium-3 as an "imminent" or "near-term" energy solution misrepresents the state of the underlying physics and engineering.

Close's argument has been published, has been responded to by Kulcinski and others, and has not in the publication's reading been refuted. The dispute is still live. The publication does not pick between Close and Kulcinski; it notes that the strongest version of the critique comes from a senior fusion physicist with no apparent conflict of interest and no commercial or programmatic stake in the answer. That is decision-relevant on its own.

The near-term market: quantum computing

The fusion case is the loud part of the helium-3 conversation. The quiet part — and the part where money is actually being committed in 2025-2026 — is the quantum-computing market.

Quantum computers based on superconducting qubits or topological qubits operate at temperatures in the low millikelvin range. Reaching those temperatures requires dilution refrigerators using helium-3. The major suppliers (Bluefors, Oxford Instruments, FormFactor) have substantial backlogs as quantum-computing companies build out infrastructure.

The quantitative scale: a single high-end research-grade dilution refrigerator uses approximately 50-200 litres of helium-3 in its operating cycle (with some recovery). Annual demand from the quantum-computing sector is in the low thousands of litres globally and growing. Current supply, from tritium-decay sources, is approximately 8,000-10,000 litres per year globally, with most of that coming from US, Russian, and (formerly) UK weapons programmes.

The supply-demand gap is real and growing. Tritium-decay production is not expanding (the source stockpiles are finite). Quantum computing is scaling. The 1,000-litre Interlune-Bluefors agreement is small in absolute terms but represents a specific market response: a commercial actor and a commercial buyer agreeing a price for a defined quantity at a defined delivery schedule.

The publication's view on this: the quantum-computing case is the case that is actually being made by money, not just by rhetoric. It is much weaker than the fusion case in long-run scale, but it is real in the near term. It would not justify the lunar industrial infrastructure that the strong fusion case is used to support. It might justify a more modest extraction-and-export operation at a small number of sites. Whether it does justify that is a question for capital allocators, not for the publication.

What the publication concludes

It does not conclude that lunar helium-3 mining will not happen. It does not conclude that fusion will or will not work. It does not adjudicate the dispute between Kulcinski and Close.

It does conclude:

1. The fusion case for lunar helium-3 mining depends on technical and economic claims that are, on the open literature, much weaker than the public framing suggests.
2. The D-D-side-reaction critique is particularly hard to answer and has not, in the publication's reading, been resolved in favour of the case for external helium-3 supply.
3. The economic case at scale closes only in best-case scenarios with very large initial investment.
4. The quantum-computing market is real, modest in scale, and the actual current driver of commercial activity.
5. The case for option-value investment in extraction capability is reasonable; the case for using "lunar helium-3 fusion will power the world" as a justification for major near-term programmes is not.

The publication's frame, named openly: the fusion claim has been a fig leaf for the resource-rush case for several decades. The resource rush may turn out to be justified on other grounds — quantum computing, water ice for propellant, option value — but those grounds need to stand on their own. Stripping away the fusion claim leaves the rest of the case to be made honestly. That is the publication's discipline.

For the broader question of why we are going at all, see the Moon public brief. For the legal framework that determines how we can extract resources, see the Treaty Framework. For the specific location where most of the early activity will happen, see the South Pole Crater Question. For the question of whether the Moon is justified as a staging point for Mars, see the Moon as Staging Point.