Helium-3 and Quantum Computing: Necessity, Supply and Scaling

Abstract

Helium-3 (³He) represents an indispensable resource for the advancement of quantum computing technologies, serving as the primary cooling medium for achieving and maintaining the ultra-low temperatures required for quantum bit (qubit) operation. This paper examines the fundamental role of ³He in dilution refrigeration systems, the physical principles underlying its necessity, current supply challenges, and implications for the quantum computing industry. We demonstrate that ³He-based cooling is not merely advantageous but essential for most contemporary quantum computing architectures, with no immediately viable alternatives at comparable performance levels. The unique quantum statistical properties of ³He, combined with its interaction with its bosonic isotope ⁴He, create a cooling mechanism that enables continuous operation at temperatures below 100 millikelvin, a regime inaccessible through any other practical means.

1. Introduction

The quantum computing revolution depends critically on maintaining quantum coherence in systems of qubits, the fundamental units of quantum information. Quantum coherence, the property that allows qubits to exist in superposition states and enables quantum algorithms to function, is extraordinarily fragile and susceptible to disruption from environmental perturbations. Thermal fluctuations represent one of the most significant sources of decoherence, as random thermal energy can excite qubits out of their quantum states and destroy the delicate superpositions upon which quantum computation depends. Most leading quantum computing platforms, including superconducting qubits, topological qubits, and certain implementations of spin qubits—require operational temperatures in the millikelvin range, typically between 10 and 100 mK, to suppress these thermal fluctuations to acceptable levels. At these temperatures, the thermal energy k_B T becomes sufficiently small relative to the energy scales of the quantum system that quantum mechanical effects dominate over classical thermal fluctuations, enabling coherence times long enough for meaningful quantum computations.

Achieving such extreme cold presents enormous technical challenges that push the boundaries of thermodynamic engineering. The temperatures required for quantum computing are more than 100 times colder than interstellar space and represent some of the coldest sustained conditions created anywhere in the universe. While liquid helium-4 (⁴He) can cool systems to approximately 1 K through evaporative cooling at reduced pressure, reaching the millikelvin regime requires fundamentally different physics. Dilution refrigeration technology, which represents the only practical method for achieving continuous cooling below 100 mK with meaningful cooling power, depends entirely on the unique properties of the rare isotope helium-3. Without ³He, the current paradigm of quantum computing based on superconducting qubits would simply not be possible.

The development of dilution refrigeration dates back to the 1960s, when the theoretical foundations were laid by Heinz London and the first working systems were constructed by Das, de Bruyn Ouboter, and Taconis in Leiden. Since then, dilution refrigerators have evolved from specialized research instruments into commercial products, though they remain among the most sophisticated cryogenic systems ever developed. The maturation of dilution refrigeration technology has occurred in parallel with the development of quantum computing, and the two fields have become inextricably linked. As quantum computing transitions from fundamental research to commercial application, the availability and cost of ³He have emerged as practical concerns that could influence the pace and scale of quantum technology deployment.

2. Fundamental Physics of Helium-3 Cooling

2.1 Quantum Statistics and the ³He-⁴He Mixture

The operation of dilution refrigerators exploits the distinct quantum statistical behaviors of ³He and ⁴He, which arise from their different nuclear compositions. Helium-4, with two protons and two neutrons for a total of four nucleons, possesses integer spin and behaves as a boson, subject to Bose-Einstein statistics. This bosonic nature allows multiple ⁴He atoms to occupy the same quantum state, leading to the remarkable phenomenon of Bose-Einstein condensation. Below its lambda point at 2.17 K, ⁴He undergoes a second-order phase transition to a superfluid state characterized by zero viscosity, the ability to flow without friction, and extraordinary thermal conductivity. This superfluid phase, known as He-II, exhibits quantum mechanical behavior on a macroscopic scale and forms the medium within which dilution refrigeration operates.

In stark contrast, ³He contains two protons but only one neutron, giving it a total of three nucleons and half-integer spin. This composition makes ³He a fermion, subject to Fermi-Dirac statistics and the Pauli exclusion principle, which forbids multiple identical fermions from occupying the same quantum state. At temperatures above approximately 2.5 mK, ³He remains a normal Fermi liquid rather than undergoing Bose-Einstein condensation. The Fermi-Dirac distribution function governs the occupancy of energy states in ³He, with a characteristic Fermi energy and Fermi temperature that depend on the density of the system. These fundamentally different quantum statistical behaviors lead to dramatically different thermodynamic properties, and it is precisely this difference that enables dilution refrigeration.

When ³He and ⁴He are mixed at temperatures below approximately 0.87 K, they spontaneously phase-separate into two distinct liquid phases in thermodynamic equilibrium. The concentrated phase contains nearly pure ³He with only a small fraction of ⁴He, while the dilute phase consists primarily of superfluid ⁴He with approximately 6.6% ³He dissolved in it at the lowest temperatures. This phase separation occurs because the zero-point energy of ³He atoms—the residual energy present even at absolute zero due to quantum mechanical uncertainty—is lower when ³He atoms are dispersed within superfluid ⁴He than when they are concentrated together in pure ³He. The equilibrium concentration of ³He in the dilute phase remains nearly constant at about 6.6% for temperatures below 100 mK, a fortunate property that simplifies the design and operation of dilution refrigerators.

The physical origin of this phase separation lies in the interplay between the zero-point motion of ³He atoms and their interactions with the surrounding medium. In the concentrated phase, ³He atoms experience strong correlations with other ³He atoms and must obey the Pauli exclusion principle, which prevents them from occupying the same quantum states. This restriction increases their average kinetic energy, as the atoms are forced into higher energy states to avoid violating the exclusion principle. In the dilute phase, ³He atoms are sufficiently separated that they can be treated approximately as independent quasi-particles moving through the superfluid ⁴He medium. The effective mass of these quasi-particles differs from the bare atomic mass due to interactions with the superfluid, but the reduced density allows the ³He atoms to occupy lower energy states on average. The chemical potential—the energy required to add one additional ³He atom to a phase—is lower in the dilute phase, driving the phase separation.

2.2 The Dilution Refrigeration Cycle

The cooling mechanism of a dilution refrigerator relies on the endothermic process of ³He atoms crossing from the concentrated phase into the dilute phase across the phase boundary. This transition requires energy because the ³He atoms must overcome an osmotic pressure difference between the two phases, analogous to the energy required for evaporation of a conventional liquid. When a ³He atom moves from the concentrated phase to the dilute phase, it must absorb energy from its surroundings to provide the enthalpy of mixing. This energy extraction from the thermal environment produces the cooling effect that makes dilution refrigeration possible. The beauty of this mechanism is that it occurs as a continuous process in thermodynamic equilibrium, with ³He atoms constantly crossing the phase boundary in both directions, but with a net flow from concentrated to dilute that can be maintained indefinitely by circulating the ³He through an external pumping system.

The thermodynamic analysis of this process reveals the fundamental temperature dependence of dilution refrigeration. The enthalpy of mixing per mole of ³He at very low temperatures is given approximately by ΔH ≈ RT²/T_F, where R is the universal gas constant, T is the operating temperature, and T_F is the Fermi temperature of ³He in the dilute phase, which is approximately 0.5 K. This quadratic temperature dependence arises from the properties of the Fermi gas, where thermal excitations at low temperature follow a T² law. The cooling power Q̇ available from a dilution refrigerator can be expressed as Q̇ ≈ 84 ṅ T² in units of watts, where ṅ is the molar circulation rate of ³He in moles per second and T is in kelvin. This fundamental relationship reveals both the power and the limitation of dilution refrigeration: the cooling power scales as the square of temperature, meaning that it decreases rapidly as temperature decreases, but it can be enhanced by increasing the circulation rate of ³He.

The practical implementation of a dilution refrigerator involves several interconnected components working in concert. The system begins with a mixing chamber where the concentrated and dilute phases coexist in equilibrium, separated by the phase boundary where the actual cooling occurs. ³He from the dilute phase is continuously removed by a pumping system, which maintains the concentration difference and drives the flow of ³He from concentrated to dilute. The removed ³He is then compressed, purified, and cooled through a series of heat exchangers before being returned to the concentrated phase to complete the cycle. The heat exchangers represent critical components, as they must cool the returning ³He from room temperature to millikelvin temperatures with minimal losses and without introducing excessive heat load. Modern dilution refrigerators employ sophisticated counterflow heat exchangers where the cold, outgoing ³He stream cools the warm, incoming stream, achieving remarkable thermal efficiency.

The still, another crucial component, operates at approximately 0.6-0.8 K and serves as the point where ³He is preferentially evaporated from the dilute phase. At these temperatures, the vapor pressure of ³He greatly exceeds that of ⁴He due to their different quantum statistics, allowing selective removal of ³He while leaving the ⁴He behind. The still must provide sufficient pumping speed to maintain the desired circulation rate while operating at a temperature accessible to conventional cryogenic refrigeration. Below the still, the ³He flows through additional heat exchangers that cool it further before it enters the mixing chamber. The thermal design of these components requires exquisite attention to detail, as even tiny heat leaks can overwhelm the limited cooling power available at millikelvin temperatures. 

Figure 1. (a,b) Modern dilution refrigerators mostly use a dry (Cryogen-free) design. They employ mechanical cryocoolers, typically Pulse Tube Cryocoolers (PTCs), to replace the Helium Bath and the 1K Pot for pre-cooling. The PTC can cool the system down to about 4K, providing the necessary cooling platform for the dilution refrigeration cycle. The main advantage of a dry system is that it eliminates the consumption of expensive and hard-to-handle liquid helium, making operations more automated and convenient. However, the mechanical motion of the PTC can introduce vibrations, requiring additional vibration-damping measures to protect sensitive quantum experiments.

2.3 Why Helium-3 is Irreplaceable

Several interconnected factors make ³He uniquely suited for ultra-low temperature refrigeration, with no known substitute offering comparable performance. The quantum degeneracy temperature of ³He, determined by its low atomic mass and fermionic nature, creates Fermi energy scales that are ideally matched to the temperature range required for quantum computing. The Fermi temperature of approximately 0.5 K in the dilute phase means that quantum degeneracy effects become significant precisely in the temperature range where cooling is needed, enabling the enthalpy of mixing that drives refrigeration. Lighter fermions would have higher Fermi temperatures and would become degenerate at impractically high temperatures, while heavier fermions would have lower Fermi temperatures and would provide insufficient cooling power in the millikelvin range.

The chemical inertness of helium provides another irreplaceable advantage. Helium is a noble gas that does not form chemical bonds under any practical conditions, ensuring that it will not react with the materials of the refrigerator or with the quantum computing components being cooled. This inertness prevents contamination, corrosion, and the formation of insulating layers that would impede thermal contact. Furthermore, helium does not condense or adsorb onto surfaces at the temperatures and pressures encountered in dilution refrigerators, avoiding complications that would arise with any reactive or condensable substance. The ability to maintain clean surfaces is particularly critical for quantum computing applications, where even atomically thin layers of contamination can introduce noise and decoherence.

The superfluid properties of ⁴He prove essential for the practical operation of dilution refrigerators. The zero viscosity of superfluid ⁴He allows mass transport without dissipative losses, enabling the circulation of the ³He-⁴He mixture through narrow channels and porous heat exchangers without generating heat. The extraordinary thermal conductivity of superfluid ⁴He, many orders of magnitude greater than that of normal liquids, ensures rapid thermal equilibration and efficient heat transfer. These properties allow the dilute phase to maintain thermal uniformity despite the ongoing cooling process and enable the design of compact, high-performance heat exchangers. No other known substance exhibits comparable superfluid properties at accessible temperatures.

Figure 2. Both ³He and ⁴He can enter a superfluid phase, however their statistical properties and superfluid mechanisms are fundamentally different as shown above.

The specific thermodynamic properties of the ³He-⁴He phase diagram create a uniquely effective cooling mechanism. The near-constant concentration of ³He in the dilute phase across a wide temperature range simplifies operation and provides predictable performance. The substantial difference in vapor pressure between ³He and ⁴He at still temperatures enables selective pumping of ³He, a feature that would be difficult or impossible to replicate with other isotopic or chemical mixtures. The enthalpies of mixing provide substantial cooling power while requiring only modest circulation rates, making the systems practical to build and operate. Attempts to find alternative cooling mechanisms using other quantum fluids, such as mixtures of different isotopes of hydrogen or other light elements, have failed to identify any system with properties approaching those of the ³He-⁴He system.

The combination of these factors, appropriate quantum degeneracy temperatures, chemical inertness, superfluid properties, and favorable thermodynamics, makes ³He not merely convenient but genuinely essential for achieving millikelvin temperatures with practical cooling powers. While alternative cooling methods exist for specific applications, none can match the combination of continuous operation, substantial cooling power, and reliability required for quantum computing. The laws of quantum mechanics and thermodynamics conspire to make ³He uniquely valuable, transforming what might seem like an arbitrary choice of coolant into a fundamental necessity determined by the nature of physical law.

3. Application in Quantum Computing Systems

3.1 Superconducting Quantum Computers

Superconducting qubits, employed by major quantum computing efforts at companies including IBM, Google, Rigetti, IQM, and numerous others, represent the most mature and widely deployed quantum computing platform currently available. These qubits are based on Josephson junctions, which are superconducting circuits containing a thin insulating barrier between two superconducting electrodes. When cooled below their critical temperature, these circuits exhibit macroscopic quantum behavior, with electrical current and voltage becoming quantum mechanical observables that can exist in superposition states. The quantum states of these circuits, typically defined by the number of Cooper pairs or the phase difference across the junction, serve as the computational basis states for quantum information processing.

The operational temperature requirements for superconducting qubits are determined by the fundamental physics of quantum coherence and thermal excitation. For a superconducting qubit with a transition frequency ω between its ground and excited states, the probability of thermal excitation is proportional to the Boltzmann factor exp(-ℏω/k_BT), where ℏ is the reduced Planck constant, k_B is Boltzmann’s constant, and T is temperature. For typical superconducting qubits operating at frequencies of 4-8 GHz, corresponding to photon energies of approximately 200-400 mK in temperature units, the thermal energy must be much smaller than the photon energy to prevent significant thermal population of excited states. The requirement that k_BT ≪ ℏω typically necessitates temperatures at least 10-20 times lower than the energy splitting, placing superconducting qubits firmly in the 10-20 mK operating range.

Figure 3. Rigetti computing’s superconducting-qubit quantum architectures. Shell: When the quantum computer is powered on, five shells (such as the white one displayed at the very top of this figure) surround it. These cans nest within each other and serve as thermal shields, keeping things inside extremely cold and vacuum-sealed. Nerves: These photon-carrying wires convey signals to and from the chip, allowing qubit operations to be driven and the measured results to be returned. Heart: The “mixing chamber” is located beneath the heat exchangers. Inside, distinct types of liquid helium (³He and ⁴He) separate and evaporate, dissipating heat. Skeleton: These gold plates serve as a barrier between cooling zones. At the bottom, the temperature drops to one hundredth of a Kelvin, which is hundreds of times colder than outer space. Brain: The QPU has a silicon chip inside a gold-plated copper disc that houses the machine’s brain, which serves as the device’s central processing unit. (Image source Rigetti computing)

Operating at these ultra-low temperatures provides multiple advantages beyond simply reducing thermal excitation. Lower temperatures reduce the density of thermally excited quasiparticles in the superconducting electrodes, which can cause decoherence by interacting with the qubit states. The quality factor of superconducting resonators, which often serve as quantum buses for coupling qubits or as readout cavities, improves at lower temperatures due to reduced losses in the superconducting materials. Lower temperatures also reduce Johnson noise from resistive elements and minimize heating effects from control electronics. The cumulative effect of these improvements is that coherence times—the duration over which quantum information remains intact—increase dramatically as temperature decreases, enabling more complex quantum algorithms with larger numbers of operations.

A typical commercial dilution refrigerator for quantum computing applications contains several grams to tens of grams of ³He, depending on the size and cooling power requirements of the system. Smaller research-scale systems might operate with 5-10 grams of ³He, providing cooling powers of 100-400 microwatts at 100 mK, sufficient for small quantum processors with tens of qubits. Larger systems designed for quantum computers with hundreds of qubits may contain 20-50 grams of ³He and provide several milliwatts of cooling power at 100 mK, enabling them to handle the increased heat load from additional qubits, control lines, and readout systems. The most advanced quantum computing facilities may employ multiple dilution refrigerators operating in parallel, each cooled by its own ³He charge, to accommodate distributed quantum processors or to provide redundancy and flexibility in experimental configurations.

The heat load on these systems arises from multiple sources that must be carefully managed. Each qubit requires multiple control lines carrying microwave signals from room temperature electronics to the millikelvin stage, and these lines conduct heat despite extensive filtering and attenuation along their length. The microwave signals themselves, when absorbed by the quantum system, deposit energy that must be removed by the refrigerator. Readout systems, which measure the quantum states of qubits, typically involve amplifiers that generate heat even when implemented with ultra-low-noise superconducting components. Residual radioactivity from materials in the environment can create quasiparticle excitations in superconducting components, contributing additional heating. The cumulative heat load from a quantum computer with hundreds of qubits can easily reach hundreds of microwatts to several milliwatts, placing substantial demands on the dilution refrigerator’s capacity.

3.2 Other Quantum Computing Platforms

Beyond superconducting qubits, several other quantum computing architectures rely on dilution refrigeration and ³He cooling to achieve their operational requirements. Topological qubits, the approach being pursued most notably by Microsoft’s quantum computing initiative, are based on exotic quasiparticles called Majorana zero modes that emerge in certain carefully engineered condensed matter systems. These systems typically involve combinations of superconducting materials, semiconductors with strong spin-orbit coupling, and precisely controlled magnetic fields. The Majorana zero modes are predicted to be topologically protected against certain types of errors, potentially providing inherently more robust qubits than conventional approaches. However, creating and maintaining the delicate conditions necessary for Majorana modes to exist requires temperatures in the 10-50 mK range, similar to or even lower than those needed for superconducting qubits. The topological protection offered by these states does not eliminate the need for ultra-low temperatures; rather, the temperatures must be low enough to ensure that the system remains in the topological phase and that thermal fluctuations do not destroy the topological properties.

Figure 4. Spectroscopy of a pair of Majorana zero modes in atomic Pb monolayer. a Schematic view of a pair of vortices in a topological superconductor, each carrying a single Majorana zero mode. b Schematic view of a single vortex whose Majorana zero mode has to be paired with a Majorana zero mode surrounding the topological domain. c Schematic structure of the Pb/Co/Si(111) samples: a Co–Si magnetic domain is buried below a Pb monolayer. d Scanning tunneling microscopy image (scale bar is 20 nm) of the sample showing a Pb monolayer with devil’s staircase structure (VT = 50 mV, IT = 50 pA). The underlying Co–Si island does not appear in the topography. e Conductance map (scale bar is 20 nm) at VT = 0 mV showing a domain with a strong zero-bias peak dot (in red) surrounded by a gapped region (dark blue) itself surrounded by a zero-bias rim (light blue). f Conductance spectra along the linecut marked in e showing a very strong peak within the gap surrounded by a gapped area and another zero-bias peak (blue) outside of the domain. g Conductance curve taken at the center of the domain showing a very high zero-bias peak (Red curve). The dashed black curve is a fit with a state at E ≈ 6 μV with an electronic temperature of 350 mK. This can be considered as a zero-energy state within the precision of the experimental set-up. h The light blue conductance curve is taken on the domain rim (light blue in e) shows a peak at zero bias. The dark blue conductance curve is taken inside the dark blue region inside the domain, it shows a hard gap with no peak at zero bias. The green and purple markers in e and d indicate the position at which those spectra were taken.

Spin qubits implemented in semiconductor quantum dots represent another important platform that benefits substantially from dilution refrigeration. These qubits encode quantum information in the spin states of individual electrons or holes confined within nanoscale regions of semiconductor materials, typically silicon or gallium arsenide. While the energy splitting between spin states can be made relatively large through the application of external magnetic fields, reaching values equivalent to several kelvin, the fidelity of quantum operations and the initialization of qubits into their ground states improve dramatically at lower temperatures. At temperatures around 1 K, achievable with pumped helium-4 systems, spin qubits can operate but with limited fidelity and initialization efficiency. When cooled to 100 mK or below using dilution refrigeration, the same qubits exhibit significantly improved performance, with higher fidelity gates and faster, more reliable initialization. Many research groups working on spin qubits therefore utilize dilution refrigeration even though it is not strictly necessary, because the enhanced performance justifies the additional complexity and cost.

The coupling between spin qubits and their electromagnetic environment provides another motivation for ultra-low temperatures. Spin qubits are typically controlled and read out using nearby charge sensors or by coupling to microwave resonators, and these ancillary systems benefit from reduced thermal noise at millikelvin temperatures. The charge sensors, often implemented as quantum point contacts or single-electron transistors, achieve better signal-to-noise ratios at lower temperatures, enabling faster and more accurate readout of qubit states. Microwave resonators used for dispersive readout of spin qubits have higher quality factors at lower temperatures, improving readout fidelity. The overall system performance, encompassing initialization, gate operations, and readout, often improves sufficiently at dilution refrigerator temperatures to justify their use even when not absolutely required.

Neutral atom qubits and trapped ion qubits, which represent alternative approaches to quantum computing, generally operate at substantially higher temperatures and do not require dilution refrigeration. Neutral atom platforms use arrays of individual atoms trapped in optical lattices or tweezers, with quantum information encoded in atomic hyperfine or Rydberg states. These systems typically operate at temperatures around 100 microkelvin, which are achieved through sophisticated laser cooling techniques rather than cryogenic refrigeration. Trapped ion systems similarly use laser cooling to reach similar temperatures, with qubits encoded in the electronic states of individual ions held in electromagnetic traps. While these temperatures are technically lower than dilution refrigerator temperatures when measured in the atoms’ motional degrees of freedom, the surrounding apparatus operates at room temperature or moderate cryogenic temperatures that do not require ³He. However, both trapped ion and neutral atom platforms face their own substantial technical challenges related to laser control, trap stability, and scaling to large numbers of qubits.

3.3 Scaling Challenges Due to ³He Shortage

As quantum computers scale from current systems with tens to hundreds of qubits toward the thousands to millions of qubits envisioned for fault-tolerant quantum computation, the cooling requirements intensify in ways that pose significant engineering challenges. Each additional qubit contributes heat load through multiple mechanisms: the control lines carrying signals to the qubit, the readout systems monitoring its state, and the dissipation from the qubit itself and its immediate environment. While a single qubit might add only a few microwatts of heat load, multiplying this by thousands of qubits creates a cumulative burden that can easily exceed the capacity of standard dilution refrigerators. The scaling is not simply linear, as larger quantum processors also require more complex control architectures, more extensive wiring, and more sophisticated multiplexing schemes, each adding its own thermal overhead.

Current state-of-the-art commercial dilution refrigerators designed for quantum computing applications typically provide cooling powers ranging from tens to hundreds of microwatts at 100 mK, with the largest systems reaching several milliwatts. These specifications represent remarkable engineering achievements, as extracting even a single milliwatt of heat at 100 mK requires circulating substantial quantities of ³He and managing heat loads at every temperature stage from room temperature down to the mixing chamber. The cooling power decreases rapidly with temperature, so a refrigerator that can remove several milliwatts at 100 mK might provide only a few hundred microwatts at 20 mK, where many superconducting qubits operate most effectively. This temperature dependence forces difficult trade-offs between operating temperature, cooling capacity, and system performance.

Next-generation quantum computers targeting thousands or tens of thousands of qubits will require substantially greater cooling capacity than currently available systems can provide. Several approaches are being pursued to meet these demands, each with its own advantages and challenges. Scaling up individual dilution refrigerators by increasing ³He circulation rates can provide more cooling power, but eventually encounters practical limits related to heat exchanger design, pressure drop in circulation loops, and the maximum flow rates achievable with available pumps. The T² dependence of cooling power means that increasing circulation rates becomes increasingly costly in terms of the required pumping infrastructure and the diminishing returns at very low temperatures. Alternative architectures using multiple dilution refrigerators in parallel can distribute the cooling load, but at the cost of increased system complexity and ³He inventory.

Distributed cooling architectures, where different sections of a large quantum processor are cooled by separate refrigerators, offer potential solutions but introduce new challenges related to synchronization and quantum interconnects between separately cooled modules. The quantum coherence must be maintained across the boundaries between different refrigerators, requiring careful engineering of electromagnetic shielding and thermal isolation while still allowing quantum information to flow between modules. Cryogenic control electronics placed at intermediate temperature stages can reduce the heat load reaching the coldest stages by performing some signal processing and multiplexing at temperatures of 4 K or higher, where cooling power is much more abundant. Superconducting microwave components, operating at these intermediate temperatures, can amplify and route signals with minimal added noise and power dissipation compared to room-temperature equivalents.

The thermal engineering of large-scale quantum computers extends beyond the dilution refrigerator itself to encompass the entire cryogenic system. Radiation shields must prevent thermal radiation from warmer stages from reaching the coldest components, as the Stefan-Boltzmann law makes radiative heat transfer increasingly important at higher temperatures. Careful attention to materials selection helps minimize heat capacity and parasitic heat loads, with every component from mounting hardware to wiring chosen for its thermal properties. Vibration isolation prevents mechanical vibrations from coupling into the quantum system and causing decoherence, while magnetic shielding attenuates external magnetic field fluctuations that could perturb qubit states. The integration of all these systems while maintaining the thermal performance necessary for quantum operation represents one of the grand engineering challenges of quantum computing.

4. Helium-3 Supply and Availability

4.1 Sources and Production

Helium-3 is extraordinarily rare on earth, with an isotopic abundance in terrestrial helium of only approximately 0.00014% or 1.4 parts per million. This extreme rarity arises from the distinct cosmological and geological origins of the two helium isotopes. Helium-4 is produced continuously through alpha decay of heavy radioactive elements in Earth’s crust and mantle, with uranium and thorium decay chains generating new ⁴He that accumulates in natural gas deposits over geological timescales. In contrast, ³He is primarily a primordial isotope, formed during Big Bang nucleosynthesis in the early universe, with only minimal terrestrial production through cosmic ray interactions and the decay of tritium. The small amount of primordial ³He that was present when Earth formed has largely escaped to space over the planet’s 4.5-billion-year history, as helium’s low atomic mass and chemical inertness allow it to reach escape velocity from Earth’s gravitational field and be lost from the atmosphere.

The primary anthropogenic source of ³He is the radioactive decay of tritium (³H), the heaviest isotope of hydrogen. Tritium undergoes beta decay with a half-life of 12.32 years according to the nuclear reaction ³H → ³He + e⁻ + ν̄_e, where an electron and an electron antineutrino are emitted along with the ³He nucleus. This decay process is the foundation of essentially all practical ³He production, as tritium can be manufactured in nuclear reactors through neutron capture reactions with lithium-6 (⁶Li + n → ⁴He + ³H) or deuterium (²H + n → ³H + γ). Historically, large quantities of tritium were produced and stockpiled by nuclear weapons programs during the Cold War, as tritium serves as a key component in boosted fission weapons and thermonuclear devices. These tritium stockpiles, maintained in specialized reservoirs, continuously generated ³He as a decay product, and this ³He was periodically extracted and made available for scientific and commercial applications.

The end of the Cold War brought dramatic reductions in tritium production as nuclear arsenals were scaled back and weapons programs downsized. In the United States, the last dedicated tritium production reactor, located at the Savannah River Site in South Carolina, ceased operations in 1988, ending domestic tritium production until the Tritium Extraction Facility came online in the 2000s to support stockpile maintenance at reduced levels. The reduction in tritium production and stockpiling had a direct impact on ³He availability, as the inventory of aging tritium in weapons reservoirs became the primary source of extracted ³He. As tritium decays exponentially with its 12.32-year half-life, the ³He production rate from a fixed tritium inventory also declines exponentially over time. A kilogram of pure tritium generates approximately 4.25 liters (at standard temperature and pressure) of ³He per year initially, but this rate decreases by about 5.5% annually due to the progressive depletion of the tritium.

Alternative sources of ³He exist but contribute relatively minor amounts to global supply. The CANDU (Canada Deuterium Uranium) reactor design, which uses heavy water (D₂O) as both moderator and coolant, produces small quantities of tritium through neutron activation of the deuterium: ²H + n → ³H + γ. This tritium dissolves in the heavy water moderator and can be extracted through specialized tritium removal facilities. The ³He subsequently produced from this tritium represents a modest but steady contribution to global supply, with Canadian production estimated at hundreds of liters per year. Some natural gas deposits contain slightly elevated ³He concentrations, particularly those in regions with high uranium and thorium content in surrounding rocks, where terrestrial nuclear processes have produced local enrichments. However, extracting ³He from natural gas is economically challenging due to the very low concentrations, and this source contributes negligibly to current supply.

Figure 5.

Helium-3 reserve and disbursements under high-and low-demand scenarios. The precipitous “cliff” in He-3 stockpile in 2008-2009 alerted officials to an impending crisis. Historical data were taken from ref. [1], which attributed them to Steve Fetter, U.S. Office of Science and Technology Policy, “Overview of helium-3 Supply and Demand,” presentation at the American Association for the Advancement of Science Workshop on helium-3, April 6, 2010. Projections were provided by Jehanne Gillo, U.S. Department of Energy (personal communication). 

The Russian Federation maintains tritium production capabilities and has historically been a significant supplier of ³He to international markets, though production levels and strategic reserves remain less transparent than Western sources. The exact scale of Russian tritium stockpiles and ³He production capacity is not publicly disclosed, but Russia has participated substantially in global ³He markets. Geopolitical considerations can affect the availability of Russian ³He to Western purchasers, introducing an element of supply uncertainty. Research reactors at various national laboratories and universities worldwide produce small amounts of tritium and subsequently ³He, but these quantities are generally consumed locally for research purposes rather than contributing to commercial markets.

The most probable and realistic source of ³He at scale is the lunar surface. ³He on the Moon originates primarily from the solar wind, a byproduct of our sun’s nuclear fusion reaction producing a continuous stream of charged particles (mostly protons and electrons, but also heavier ions including helium nuclei) that flows outward at speeds of 300 to 800 kilometers per second. Unlike Earth, which is shielded by a global magnetic field and thick atmosphere that deflect and absorb these particles, the Moon has no such protection. Over billions of years, solar wind ions have bombarded the lunar surface directly, implanting themselves into the top layers of the regolith. ³He accumulates in the fine grains of lunar soil, adhering to mineral surfaces and becoming trapped within the glassy agglutinates within the regolith. Conservative estimates suggest the Moon holds at least 1.1 million metric tons of ³He in its upper regolith. Companies such as Magna Petra are leading the charge in lunar isotope extraction technology with prospecting missions planned for 2027 and extraction/return missions planned for 2029/2030.

4.2 Supply Crisis and Market Dynamics

The ³He shortage emerged dramatically in the late 2000s when demand surged while supply remained constrained by the fundamental limitations of tritium decay inventory. The immediate trigger was the expansion of neutron detection capabilities for homeland security applications following the September 11, 2001 terrorist attacks and subsequent concerns about nuclear terrorism. Helium-3 serves as an excellent neutron detector through the nuclear reaction ³He + n → ³H + ¹H + 764 keV, which has an extremely large cross-section for thermal neutrons and produces easily detected charged particles. The U.S. Department of Homeland Security initiated a massive program to deploy radiation portal monitors at ports, border crossings, and other strategic locations to detect potential smuggling of nuclear materials. Each portal monitor required significant quantities of ³He-filled proportional counters, and the rapid scaling of this program created demand that rapidly overwhelmed available supply.

Figure 6. “Caught by Surprise” Helium-3 crisis, highlighted in 2010 congressional hearings, stemmed from dwindling production (a byproduct of tritium for nuclear arms) colliding with soaring demand for neutron detectors in homeland security post-9/11, depleting the US stockpile, causing price spikes, and revealing supply chain vulnerabilities, prompting research into alternatives and long-term strategies. 

The supply-demand imbalance manifested first as allocation restrictions and lengthening delivery times, then as dramatic price increases as suppliers attempted to ration limited stocks. Prior to the shortage, ³He traded at approximately $100-200 per liter at standard temperature and pressure (STP), measured at 273 K and 101.325 kPa pressure. By 2008-2009, prices had escalated to $2,000-3,000 per liter with current (2026) pricing at between $6,000-$15,000 per liter, representing a continual increase year over year. The rapid price escalation creates severe budgetary challenges for research programs and commercial applications that depended on ³He, forcing difficult decisions about project continuation and technology alternatives.

The shortage prompted immediate responses from both suppliers and consumers. The U.S. Department of Energy, which controls the primary domestic ³He stockpile derived from weapons program tritium, implemented formal allocation procedures to prioritize critical national security applications while attempting to maintain support for scientific research. Many research projects faced delays or cancellations as ³He became unavailable at any price, and waiting lists extended for months or years. The scientific community responded by exploring alternative detector technologies for neutron detection, including boron-lined proportional counters, lithium-loaded glass scintillators, and various solid-state detectors. While none of these alternatives matched the performance of ³He detectors in all respects, they proved adequate for many applications and reduced pressure on ³He supplies.

Current production derives from multiple sources: the U.S. Department of Energy’s existing tritium inventory, which continues to decay and produce ³He albeit at declining rates; Canadian heavy water reactor operations; Russian sources of uncertain magnitude; and small contributions from various research and commercial facilities worldwide. The rate of ³He generation from any fixed tritium inventory decreases exponentially over time, meaning that without new tritium production or stockpiling, the decay-derived ³He supply will continue to decline. The U.S. has resumed limited tritium production to maintain its nuclear stockpile, but at levels far below Cold War peaks, and most of this tritium is needed for weapons maintenance rather than ³He generation.

The supply for ³He remains relatively small and specialized compared to commodity gases, with total annual production measured in thousands of liters rather than millions of cubic meters. This limited market size means few suppliers exist, and those that do operate often function as intermediaries between government-controlled stockpiles and end users rather than as independent producers. Market prices remain elevated and subject to fluctuation based on allocation policies and demand from various sectors. Long-term supply contracts and advance planning have become essential for large-scale users, as spot market availability can be limited or nonexistent. The strategic importance of ³He has led some countries to classify supply data and restrict exports, further complicating market dynamics.

4.3 Implications for Quantum Computing

The quantum computing industry currently represents a fraction of total ³He demand compared to established applications such as neutron detection, cryogenics research, magnetic resonance imaging, and fundamental physics experiments. However, the growth trajectory of quantum computing suggests that this balance could shift substantially over the coming decade. Each dilution refrigerator for quantum computing contains anywhere from several grams to several tens of grams of ³He, depending on system size and design. To convert to more familiar units, one gram of ³He occupies approximately 7.5 liters at standard temperature and pressure, meaning a typical quantum computing refrigerator might contain 40-400 liters STP of ³He. At current market prices of $15,000-17,000 per liter, this translates to $600,000-$6,800,000 worth of ³He per refrigerator, representing a substantial capital investment before even considering the refrigerator hardware and infrastructure.

As quantum computing transitions from research laboratories to commercial deployment, the number of operational quantum computers is growing rapidly. Companies including IBM, Google, Amazon, Microsoft, and numerous startups are operating or developing quantum computing systems, each requiring dilution refrigeration. Research institutions, universities, and government laboratories worldwide contribute additional systems. If the quantum computing industry follows optimistic growth projections, with hundreds to thousands of systems deployed globally by 2030, the cumulative ³He need will reach tens to hundreds of thousands of liters STP. This would represent a substantial fraction of annual global production, transforming quantum computing from a negligible consumer of ³He to a major demand driver competing with other critical applications.

The competition for ³He extends across multiple sectors with strong claims to priority. National security applications, particularly neutron detection for nuclear nonproliferation and counterterrorism, carry obvious importance for government allocation decisions. Fundamental physics research, including experiments in particle physics and cosmology that rely on ultra-low temperature techniques, represents irreplaceable knowledge generation with no alternative approaches in many cases. Medical applications, such as ³He magnetic resonance imaging for lung ventilation studies, provide unique diagnostic capabilities for respiratory diseases including chronic obstructive pulmonary disease, asthma, and cystic fibrosis. Industrial applications in cryogenics and materials science support various technological developments. Each of these sectors can make persuasive arguments for access to limited ³He supplies, creating complex trade-offs for allocation authorities.

The strategic implications of ³He dependence have not been lost on quantum computing companies and national technology initiatives. Recognition of supply constraints has driven several responses. First, refrigerator manufacturers have focused intensively on ³He conservation, developing systems with minimal leakage rates, improved recovery capabilities during maintenance operations, and designs that minimize the required ³He inventory while maintaining adequate cooling power. Modern dilution refrigerators achieve leak rates of less than 0.1% of inventory per year under normal operation, a remarkable improvement over earlier designs. Second, companies are establishing long-term relationships with suppliers and securing advance commitments for ³He supply to ensure availability for future expansion. Third, recycling and refurbishment of ³He from decommissioned systems has become standard practice, with specialized facilities offering purification and recertification services for recovered material.

The geopolitical dimension of ³He supply adds another layer of complexity to quantum computing development. ³He production is concentrated in relatively few countries, primarily the United States, Canada, and Russia, with limited transparency about inventory levels and production capabilities. Export controls and strategic considerations can restrict international flows of ³He, particularly to countries or entities deemed security risks. Quantum computing, with its potential applications in cryptography, optimization, and simulation relevant to national security, may face particular scrutiny in ³He allocation decisions. Countries developing indigenous quantum computing capabilities must either secure domestic ³He sources, establish reliable international supply relationships, or invest in alternatives that reduce ³He dependence.

The economic structure of ³He markets creates additional challenges for quantum computing applications. The small market size and limited number of suppliers mean that prices can be volatile and supply can be subject to nonmarket allocation mechanisms. Long lead times between ordering and delivery require extensive advance planning, complicating rapid scaling of quantum computing operations. The high value and strategic nature of ³He necessitates extensive inventory tracking, security measures, and regulatory compliance, adding administrative overhead. These factors collectively increase the total cost of ownership for quantum computing systems beyond the direct ³He purchase price, affecting business models and deployment strategies.

5. Quantum Computing Scaling and Helium-3 Demand Projections

The trajectory of quantum computing development and commercialization will fundamentally determine future demand for helium-3, with growth projections spanning a wide range depending on assumptions about technological progress, market adoption, and architectural choices. Understanding these demand projections requires examining both the scaling of individual quantum computers, from current systems with tens to hundreds of qubits toward fault-tolerant machines with millions of qubits, and the proliferation of quantum computing installations globally as the technology matures from research prototypes to commercial products and services. The interplay between these two scaling dimensions, combined with the specific cooling requirements of different qubit technologies, will shape the quantum computing industry’s ³He footprint over the coming decades.

5.1 Scaling Individual Quantum Processors

The roadmaps published by leading quantum computing companies reveal ambitious plans for increasing qubit counts that will directly impact cooling requirements and ³He demand per system. IBM’s quantum computing roadmap, among the most detailed publicly available, projects scaling from their current systems with hundreds of qubits to systems with thousands of qubits in the mid-2020s, with explicit targets including a 4,000-qubit processor. Beyond these near-term milestones, the pathway to fault-tolerant quantum computing, widely considered necessary for most commercially valuable applications, requires dramatically larger systems. Fault-tolerant quantum computation relies on quantum error correction, where each logical qubit is encoded across multiple physical qubits to detect and correct errors that inevitably occur during computation. The overhead imposed by error correction varies depending on the specific error correction code and the physical error rates of the underlying qubits, but representative estimates suggest that implementing a single high-quality logical qubit might require anywhere from hundreds to thousands of physical qubits.

The quantum computing architecture community has extensively analyzed the resource requirements for running useful algorithms on fault-tolerant quantum computers. Studies examining algorithms such as Shor’s algorithm for factoring large numbers, quantum simulation of molecular systems for drug discovery and materials design, and optimization algorithms for logistics and finance consistently arrive at estimates requiring millions of physical qubits for practical problem sizes. A frequently cited analysis of breaking RSA encryption with Shor’s algorithm suggests that factoring a 2048-bit number, a standard key length in current cryptographic systems, would require approximately 20 million physical qubits assuming realistic error rates and error correction overheads. Similarly, quantum simulations of industrially relevant chemical systems, such as the nitrogenase enzyme for nitrogen fixation or complex battery cathode materials, have been estimated to require millions of qubits for chemically accurate results that could guide experimental synthesis.

These projections for fault-tolerant systems carry profound implications for cooling infrastructure. A quantum computer with one million physical superconducting qubits, each requiring control and readout lines penetrating from room temperature to millikelvin temperatures, would present heat loads far exceeding the capacity of current dilution refrigeration technology. Each qubit’s control lines contribute thermal conduction, even with extensive filtering and attenuation at intermediate temperature stages. The microwave control signals, though carefully calibrated to minimize energy deposition, still dissipate power when absorbed by the quantum system or when reflected signals are terminated in matched loads. Readout amplifiers, even when implemented with ultra-low-noise superconducting parametric amplifiers or traveling-wave parametric amplifiers, generate heat that must be removed to prevent warming the qubit environment. Conservative estimates suggest that each qubit might contribute 1-10 microwatts of heat load to the coldest stage when accounting for all these sources, implying that a million-qubit system could face total heat loads of 1-10 watts at millikelvin temperatures.

Extracting watts of heat at millikelvin temperatures vastly exceeds the capabilities of any existing dilution refrigerator by several orders of magnitude. The fundamental T² scaling of dilution refrigeration cooling power means that achieving such performance would require either operating at significantly higher temperatures, which would compromise qubit performance, or dramatically increasing ³He circulation rates to levels that pose severe engineering challenges. A more plausible architecture for million-qubit systems involves distributing qubits across multiple refrigeration modules, with each module cooled independently and quantum information transferred between modules through specialized interconnects. If each dilution refrigerator could accommodate 1,000-10,000 qubits given optimized thermal engineering, a million-qubit system might require 100-1,000 separate refrigeration modules operating in parallel. This distributed architecture introduces its own challenges related to synchronization, coherence preservation across module boundaries, and the complexity of managing such an extensive cryogenic infrastructure.

The ³He inventory required for these scaled systems depends critically on the specific refrigerator designs employed. Current commercial dilution refrigerators designed for quantum computing typically contain 5-50 grams of ³He depending on their cooling power specifications. Extrapolating to future systems requires assumptions about refrigerator efficiency improvements and architectural choices. If refrigerator technology continues to improve, enabling greater cooling power per unit of ³He through enhanced heat exchanger designs and optimized circulation systems, the ³He cost per qubit could decrease over time even as total qubit counts increase. Conversely, if scaling to higher cooling powers encounters fundamental limitations requiring proportional increases in ³He inventory, the material requirements could scale roughly linearly or even superlinearly with cooling capacity. A conservative middle-ground estimate might assume that a refrigerator capable of supporting 10,000 qubits would require 50-100 grams of ³He, implying that a million-qubit system distributed across 100 refrigerators would require 5-10 kilograms of ³He total, or approximately 37,000-75,000 liters at standard temperature and pressure.

5.2 Global Proliferation of Quantum Computing Systems

Beyond the scaling of individual quantum processors, the total demand for ³He will be determined by the number of quantum computing installations deployed worldwide. Market forecasting for emerging technologies inherently involves substantial uncertainty, as adoption rates depend on factors including technological maturation, cost reductions, demonstration of commercial value, ecosystem development, and competition from alternative approaches. Nevertheless, various market research firms, consulting companies, and industry analysts have published projections for quantum computing market size and deployment numbers that, while necessarily speculative, provide frameworks for understanding potential demand scenarios.

Market research reports examining the quantum computing industry have projected significant growth in both spending and system deployments over the next decade. While specific numerical projections vary widely depending on assumptions and methodologies, growth-oriented scenarios envision quantum computing transitioning from a research curiosity to an established technology platform with substantial commercial deployment. Some analysts project the quantum computing market reaching multi-billion dollar annual revenues by 2030, with increasing fractions of this spending directed toward hardware systems as opposed to purely research and development activities. The number of operational quantum computing systems, both on-premises installations and systems deployed in cloud quantum computing facilities, would necessarily grow to support this market expansion.

Figure 7. The quantum computing market is experiencing significant growth at a CAGR of 25.4%, with revenues projected to exponentially  surge in the coming years. -market.us

The business models emerging in quantum computing will influence deployment patterns and thus total system counts. Cloud quantum computing platforms, offered by companies including IBM, Amazon (through AWS Braket), Google, Microsoft (Azure Quantum), and others, centralize quantum processors in data center-like facilities where they are accessed remotely by users. This cloud-centric approach potentially reduces total system proliferation compared to scenarios where every organization requiring quantum computing capabilities operates its own on-premises hardware. However, as the technology matures and costs decline, on-premises systems become more attractive for organizations with sustained high-volume quantum computing needs, security concerns about cloud access, or requirements for tight integration with classical computing infrastructure. The eventual balance between cloud and on-premises deployments will substantially affect total refrigerator counts and ³He demand.

Industry roadmaps and company announcements provide some indication of near-term growth expectations. IBM has articulated a vision of making quantum computing widely accessible through its quantum network, which has grown to include more than 100 member organizations accessing IBM’s quantum systems. While many of these members currently share access to a limited number of systems, the expansion of membership and the introduction of larger, more capable systems suggest growing demand. Google’s quantum computing efforts, while more research-focused, have demonstrated increasing investment and system capabilities. Startups including Rigetti, IonQ, and numerous others have secured substantial venture capital funding premised on expectations of commercial quantum computing markets materializing over the next 5-10 years. Government quantum initiatives in the United States, European Union, China, Japan, and other countries have committed multi-billion dollar investments over multi-year periods, with portions of this funding directed toward building quantum computing infrastructure.

China’s quantum technology investments deserve particular attention given their scale and potential impact on global ³He demand. The Chinese government has identified quantum information science as a strategic priority, with investments estimated in the billions of dollars through various programs including the National Laboratory for Quantum Information Sciences in Hefei and quantum computing research centers at leading universities. While China has made notable progress in quantum communication and quantum sensing, quantum computing development has also advanced, with research groups demonstrating superconducting qubit systems and making progress toward larger processors. If China’s quantum computing efforts scale to deployment of numerous large systems, either for research purposes or for eventual commercial applications, the corresponding ³He demand could be substantial. The geopolitical dimension of ³He supply takes on added significance in this context, as Chinese access to ³He from Western sources could face restrictions, potentially driving Chinese investment in alternative ³He sources or cooling technologies.

The quantum computing industry’s growth trajectory shows particular acceleration in recent years. Venture capital investment in quantum computing companies reached record levels in 2021 and has remained elevated despite broader technology sector downturns, indicating sustained investor confidence in long-term commercial potential. Government spending on quantum information science has similarly increased globally, with the U.S. National Quantum Initiative Act authorizing substantial multi-year funding, the EU’s Quantum Flagship program directing approximately €1 billion over ten years, and comparable initiatives in other countries. This financial commitment is translating into expanded research facilities, increased hiring of specialized personnel, and growing numbers of quantum computing systems being built and operated.

5.3 Demand Scenarios and Sensitivity Analysis

Constructing plausible scenarios for future ³He demand from quantum computing requires combining assumptions about individual system scaling, global deployment numbers, and the fraction of quantum computers employing superconducting or other cryogenic qubit technologies that require dilution refrigeration. The scenarios necessarily span a wide range, reflecting genuine uncertainty about technological and market developments, but the upper bounds of plausible projections suggest ³He demand that could strain or exceed current global supply.

A conservative baseline scenario might assume relatively slow scaling of quantum computers and modest global deployment. In this scenario, quantum computing over the next decade remains primarily a research and development activity with limited commercial deployment. Individual systems might scale to thousands of qubits but not beyond, as technical challenges in error correction and system integration prove more difficult than anticipated. Global deployment might reach several hundred systems by 2030, distributed across research institutions, major technology companies, and government laboratories. Assuming an average of 30 grams of ³He per system (225 liters STP), several hundred systems would require approximately 100,000 liters STP of ³He total inventory, representing a significant fraction of current annual global production. This scenario shows quantum computing would become a main consumer of ³He, competing alongside existing applications and requiring careful allocation management.

A moderate growth scenario incorporates more optimistic assumptions about technological progress and commercial adoption while remaining plausible based on current trajectories. In this scenario, quantum computing achieves early demonstrations of quantum advantage for specific commercial applications, perhaps in optimization, machine learning, or materials simulation, by the mid-to-late 2020s, driving increased commercial interest and investment. Individual systems scale to tens of thousands of qubits as companies make progress toward fault-tolerant architectures, requiring either larger refrigerators with correspondingly more ³He or multiple refrigerators per system. Global deployment reaches 1,000-2,000 systems by 2030, including cloud platforms, corporate research facilities, and government installations. With an average of 100 grams of ³He per system (750 liters STP) accounting for larger processor sizes and possible multi-refrigerator architectures, total inventory reaches 750,000-1,500,000 liters STP.

An aggressive high-growth scenario, representing the upper bound of current industry projections, envisions rapid commercialization of quantum computing with widespread deployment across multiple sectors. In this scenario, breakthrough demonstrations of quantum advantage in commercially valuable applications occur in the mid/late-2020s, triggering rapid industry adoption similar to the growth curves observed in previous transformative computing technologies. Individual systems scale aggressively toward early fault-tolerant architectures with 100,000-1,000,000 physical qubits by 2030, distributed across multiple refrigeration modules. Global deployment accelerates dramatically, reaching 5,000-10,000 systems worldwide as cloud providers expand capacity, large corporations build private quantum computing facilities, and government and research institutions proliferate systems for various applications. If we assume these advanced systems average 500 grams of ³He each (3,750 liters STP) due to their multi-refrigerator distributed architectures, total inventory for 5,000-10,000 systems reaches 18,750,000-37,500,000 liters STP. This represents an enormous cumulative demand that would be impossible to satisfy with existing supply chains and would require transformative changes in either ³He production or quantum computing cooling approaches.

The sensitivity of these projections to key assumptions deserves careful examination. The ³He requirement per system varies substantially based on refrigerator design efficiency, with potential improvements in heat exchanger technology, circulation system design, and thermal management possibly reducing inventory needs by factors of two or more. Conversely, if larger processors require cooling powers that scale superlinearly with ³He inventory, requirements could exceed linear projections. The fraction of quantum computers using superconducting qubits versus other architectures substantially affects total demand, as trapped ion and photonic systems generally avoid ³He requirements. If neutral atom or photonic quantum computing achieves superior scaling or commercial viability compared to superconducting approaches, total industry ³He demand could remain modest even with large numbers of deployed systems. However, current technological momentum and commercial investment remain heavily weighted toward superconducting platforms, suggesting they will dominate deployments at least through the next 5-10 years.

The geographical distribution of quantum computing deployment carries implications for ³He supply dynamics beyond simple total demand. Concentrated deployment in countries with domestic ³He production capabilities (primarily the United States, Canada, and Russia) creates different supply chain dynamics than widely distributed global deployment. Countries building substantial quantum computing capabilities without domestic ³He sources must establish reliable import relationships or invest in domestic production infrastructure, potentially including tritium breeding in nuclear reactors. The strategic nature of quantum computing, with applications in cryptography, national security, and economic competitiveness, may motivate governments to treat ³He access as a technology sovereignty issue, potentially fragmenting global supply chains and complicating market dynamics.

The temporal dynamics of demand growth also matter for supply adequacy. Rapid deployment concentrated over 5 years would strain supply chains more severely, as production capacity cannot expand quickly due to the fundamental constraints of tritium decay rates and the long timelines for nuclear infrastructure development. The scenarios outlined above implicitly assume deployment over approximately a decade, but acceleration or deceleration of commercialization timelines would shift supply-demand dynamics accordingly.

Estimates of current global ³He production vary in the literature, with most authoritative sources suggesting annual production from tritium decay is approximately 8,000-15,000 liters STP, primarily derived from the U.S. Department of Energy’s aging tritium stockpiles. It is important to distinguish between production capacity and historical consumption: during peak demand in 2008, approximately 70,000-80,000 liters were consumed, but this drew down accumulated stockpiles rather than reflecting sustainable production. The decline in tritium stockpiles from their Cold War peaks means that production from decay continues to decrease over time unless offset by new tritium production. This constrained baseline production makes the quantum computing demand scenarios even more significant, as the industry would be competing for an increasingly limited resource unless substantial changes occur in either supply expansion or quantum computing cooling technology.

6. Conclusion

Helium-3 occupies a position of critical and irreplaceable importance in the contemporary quantum computing industry, enabling the ultra-low temperatures required for superconducting qubits and other cryogenic quantum computing architectures to achieve the quantum coherence necessary for computation. The unique quantum mechanical properties of ³He, arising from its fermionic nature and its interaction with bosonic ⁴He in dilution refrigeration systems, create a cooling mechanism that represents the only practical method for achieving continuous operation at millikelvin temperatures with sufficient cooling power for quantum computing applications. The physics underlying this necessity is not a matter of engineering convenience or economic optimization but rather reflects fundamental constraints imposed by quantum statistics and thermodynamics. No alternative cooling technology currently exists that can match the combination of temperature range, cooling power, reliability, and continuous operation provided by ³He-based dilution refrigeration.

Current production of ³He presents substantial challenges for the scaling quantum computing industry due to the extreme scarcity of this isotope and the limited pathways for its production. With global annual production from tritium decay estimated at 8,000-15,000 liters STP, primarily derived from the radioactive decay of aging tritium stockpiles that themselves represent a declining resource, the supply of ³He is fundamentally constrained. Historical consumption during the 2008-2009 shortage reached 70,000-80,000 liters annually, but this depleted accumulated reserves and was not sustainable. The dramatic price increases and allocation restrictions experienced during the ³He shortage demonstrate the fragility of supply-demand balance and the potential for disruption when new applications create surges in demand. The quantum computing industry currently represents a relatively modest consumer of ³He compared to established applications in neutron detection, fundamental physics research, and medical imaging, but this balance could shift dramatically as quantum computing scales from research laboratories to commercial deployment.

The implications of these projections extend beyond simple resource economics to encompass strategic and geopolitical dimensions. Quantum computing, with additional applications spanning nuclear fusion energy, cryptography, drug discovery, materials design, financial modeling, and artificial intelligence, represents a technology of enormous economic and national security significance. Countries and companies pursuing quantum computing capabilities must secure access to ³He either through domestic production, reliable international supply relationships, or investment in alternative approaches that reduce or eliminate ³He dependence. The concentration of ³He production in a small number of countries, primarily the United States, Canada, and Russia, creates potential vulnerabilities in supply chains and opportunities for strategic competition. Export controls, allocation priorities favoring national security applications, and geopolitical tensions could all restrict access to ³He for quantum computing applications, particularly for countries or entities deemed security risks.

The interplay between ³He availability and quantum computing development exemplifies the complex material dependencies that underlie cutting-edge technologies and the ways in which fundamental physical constraints can shape technological trajectories. Unlike software or information-based innovations that can scale with minimal material inputs, quantum computing hardware depends on access to specific physical resources with properties determined by the laws of quantum mechanics. The scarcity of ³He, arising from cosmological and nuclear processes operating over timescales ranging from the Big Bang to the 12.32-year half-life of tritium, imposes real constraints on the pace and architecture of quantum computing deployment. These constraints are not merely economic, to be overcome through sufficient investment, but reflect fundamental limitations in the availability of matter with the required properties.

As quantum computing continues its transition from fundamental research to commercial application and eventual widespread deployment, the ³He supply question will require ongoing attention and proactive management. Quantum computing companies must incorporate ³He availability into their technology roadmaps and business models, making strategic decisions about refrigerator procurement, inventory management, and investment in alternative cooling approaches. Governments must balance competing demands for ³He across applications including national security, scientific research, medical diagnostics, and emerging technologies, potentially developing formal allocation mechanisms or investing in supply expansion. The scientific and engineering communities must continue pursuing efficiency improvements, recycling technologies, and alternative approaches that could ease supply pressures. International cooperation, or alternatively strategic competition, over ³He access will influence which countries and companies can successfully deploy quantum computing capabilities at scale.

What remains clear is that ³He, this rare isotope produced by nuclear decay and stellar processes, has become inextricably linked to humanity’s most ambitious attempt to harness quantum mechanics for computation. The quantum computing revolution, if it fully materializes, will be built not only on advances in physics, engineering, and computer science, but also on access to adequate supplies of a substance whose scarcity reflects the deep structure of nuclear physics and cosmology. This dependence serves as a reminder that even the most advanced technologies remain grounded in the material world and subject to constraints that cannot be eliminated through cleverness or investment alone. How the quantum computing industry navigates these constraints over the coming decades will substantially influence both the technology’s trajectory and our broader understanding of the relationship between technological ambition and physical reality.

Works Cited

Dilution Refrigeration and Cryogenic Physics

1. London, H., Clarke, G. R., & Mendoza, E. (1962). Osmotic pressure of ³He in liquid ⁴He, with proposals for a refrigerator to work below 1°K. Physical Review, 128(5), 1992-2005.
2. Lounasmaa, O. V. (1974). Experimental Principles and Methods Below 1K. Academic Press.
3. Pobell, F. (2007). Matter and Methods at Low Temperatures (3rd ed.). Springer-Verlag.
4. Zu, C., Wang, W. B., He, L., Zhang, W. G., Dai, C. Y., Wang, F., & Duan, L. M. (2014). Experimental realization of universal geometric quantum gates with solid-state spins. Nature, 514(7520), 72-75.
5. Frossati, G. (1978). Experimental techniques: Methods for cooling below 300 mK. Journal de Physique Colloques, 39(C6), C6-1578-C6-1589.
6. Radebaugh, R. (2009). Cryocoolers: The state of the art and recent developments. Journal of Physics: Condensed Matter, 21(16), 164219.
7. Richardson, R. C., & Smith, E. N. (Eds.). (1988). Experimental Techniques in Condensed Matter Physics at Low Temperatures. Addison-Wesley.

Quantum Computing Hardware and Architecture

8. Devoret, M. H., & Schoelkopf, R. J. (2013). Superconducting circuits for quantum information: An outlook. Science, 339(6124), 1169-1174.
9. Koch, J., Yu, T. M., Gambetta, J., Houck, A. A., Schuster, D. I., Majer, J., … & Schoelkopf, R. J. (2007). Charge-insensitive qubit design derived from the Cooper pair box. Physical Review A, 76(4), 042319.
10. Barends, R., Kelly, J., Megrant, A., Veitia, A., Sank, D., Jeffrey, E., … & Martinis, J. M. (2014). Superconducting quantum circuits at the surface code threshold for fault tolerance. Nature, 508(7497), 500-503.
11. Arute, F., Arya, K., Babbush, R., Bacon, D., Bardin, J. C., Barends, R., … & Martinis, J. M. (2019). Quantum supremacy using a programmable superconducting processor. Nature, 574(7779), 505-510.
12. Krinner, S., Lacroix, N., Remm, A., Di Paolo, A., Genois, E., Leroux, C., … & Eichler, C. (2022). Realizing repeated quantum error correction in a distance-three surface code. Nature, 605(7911), 669-674.
13. Loss, D., & DiVincenzo, D. P. (1998). Quantum computation with quantum dots. Physical Review A, 57(1), 120-126.
14. Veldhorst, M., Yang, C. H., Hwang, J. C., Huang, W., Dehollain, J. P., Muhonen, J. T., … & Dzurak, A. S. (2015). A two-qubit logic gate in silicon. Nature, 526(7573), 410-414.
15. Nayak, C., Simon, S. H., Stern, A., Freedman, M., & Das Sarma, S. (2008). Non-Abelian anyons and topological quantum computation. Reviews of Modern Physics, 80(3), 1083-1159.

Quantum Error Correction and Fault Tolerance

16. Fowler, A. G., Mariantoni, M., Martinis, J. M., & Cleland, A. N. (2012). Surface codes: Towards practical large-scale quantum computation. Physical Review A, 86(3), 032324.
17. Gidney, C., & Ekerå, M. (2021). How to factor 2048 bit RSA integers in 8 hours using 20 million noisy qubits. Quantum, 5, 433.
18. Reiher, M., Wiebe, N., Svore, K. M., Wecker, D., & Troyer, M. (2017). Elucidating reaction mechanisms on quantum computers. Proceedings of the National Academy of Sciences, 114(29), 7555-7560.
19. Berry, D. W., Gidney, C., Motta, M., McClean, J. R., & Babbush, R. (2019). Qubitization of arbitrary basis quantum chemistry leveraging sparsity and low rank factorization. Quantum, 3, 208.
20. Campbell, E. T., Terhal, B. M., & Vuillot, C. (2017). Roads towards fault-tolerant universal quantum computation. Nature, 549(7671), 172-179.

Helium-3 Supply and Production

21. Shea, D. A., & Morgan, D. (2010). The Helium-3 Shortage: Supply, Demand, and Options for Congress (CRS Report No. R41419). Congressional Research Service.
22. National Research Council. (2012). Selling the Nation’s Helium Reserve. National Academies Press.
23. U.S. Department of Energy. (2009). Helium-3 shortage could cripple national security and research. DOE/NSF Nuclear Science Advisory Committee Report.
24. Kouzes, R. T., Ely, J. H., Erikson, L. E., Kernan, W. J., Lintereur, A. T., Siciliano, E. R., … & Warren, G. A. (2010). Neutron detection alternatives to ³He for national security applications. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 623(3), 1035-1045.
25. Anderson, J. L., Valentine, J. D., Morichi, M., & Azmy, Y. (2010). Examination of neutron detection alternatives to helium-3. Oak Ridge National Laboratory Report ORNL/TM-2010/207.
26. Van Sciver, S. W. (2012). Helium cryogenics. In International Cryogenics Monograph Series (2nd ed.). Springer.
27. Cho, A. (2009). Helium-3 shortage could put freeze on low-temperature research. Science, 326(5954), 778-779.

Quantum Computing Industry and Market Analysis

28. Boston Consulting Group. (2021). The Next Decade in Quantum Computing—and How to Play. BCG Henderson Institute.
29. McKinsey & Company. (2021). Quantum computing: An emerging ecosystem and industry use cases. McKinsey Digital.
30. IBM Quantum. (2023). IBM Quantum Development Roadmap. Retrieved from https://www.ibm.com/quantum/roadmap
31. Gambetta, J. M., Chow, J. M., & Steffen, M. (2017). Building logical qubits in a superconducting quantum computing system. npj Quantum Information, 3(1), 2.

Government Quantum Initiatives

32. National Science and Technology Council. (2018). National Strategic Overview for Quantum Information Science. Executive Office of the President.
33. U.S. Congress. (2018). National Quantum Initiative Act (Public Law 115-368). 115th Congress.
34. European Commission. (2018). Quantum Technologies Flagship. Directorate-General for Communications Networks, Content and Technology.
35. Quantum Economic Development Consortium. (2021). The Quantum Industry: A U.S. Perspective. QED-C Technical Report.

Materials Science and Qubit Development

36. Clarke, J., & Wilhelm, F. K. (2008). Superconducting quantum bits. Nature, 453(7198), 1031-1042.
37. Kjaergaard, M., Schwartz, M. E., Braumüller, J., Krantz, P., Wang, J. I. J., Gustavsson, S., & Oliver, W. D. (2020). Superconducting qubits: Current state of play. Annual Review of Condensed Matter Physics, 11, 369-395.
38. Reagor, M., Paik, H., Catelani, G., Sun, L., Axline, C., Holland, E., … & Schoelkopf, R. J. (2016). Reaching 10 ms single photon lifetimes for superconducting aluminum cavities. Applied Physics Letters, 102(19), 192604.
39. Pop, I. M., Geerlings, K., Catelani, G., Schoelkopf, R. J., Glazman, L. I., & Devoret, M. H. (2014). Coherent suppression of electromagnetic dissipation due to superconducting quasiparticles. Nature, 508(7496), 369-372.

Thermodynamics and Statistical Mechanics

40. Wilks, J. (1967). The Properties of Liquid and Solid Helium. Clarendon Press.
41. Enss, C., & Hunklinger, S. (2005). Low-Temperature Physics. Springer-Verlag.
42. Landau, L. D. (1941). The theory of superfluidity of helium II. Journal of Physics USSR, 5, 71-90.
43. Bardeen, J., Baym, G., & Pines, D. (1967). Effective interaction of ³He atoms in dilute solutions of ³He in ⁴He at low temperatures. Physical Review, 156(1), 207-221.

Alternative Quantum Computing Platforms

44. Bruzewicz, C. D., Chiaverini, J., McConnell, R., & Sage, J. M. (2019). Trapped-ion quantum computing: Progress and challenges. Applied Physics Reviews, 6(2), 021314.
45. Bluvstein, D., Levine, H., Semeghini, G., Wang, T. T., Ebadi, S., Kalinowski, M., … & Lukin, M. D. (2022). A quantum processor based on coherent transport of entangled atom arrays. Nature, 604(7906), 451-456.
46. Flamini, F., Spagnolo, N., & Sciarrino, F. (2019). Photonic quantum information processing: A review. Reports on Progress in Physics, 82(1), 016001.
47. Madsen, L. S., Laudenbach, F., Askarani, M. F., Rortais, F., Vincent, T., Bulmer, J. F., … & Laing, A. (2022). Quantum computational advantage with a programmable photonic processor. Nature, 606(7912), 75-81.

Cryogenic Engineering for Quantum Computing

48. Krinner, S., Storz, S., Kurpiers, P., Magnard, P., Heinsoo, J., Keller, R., … & Wallraff, A. (2019). Engineering cryogenic setups for 100-qubit scale superconducting circuit systems. EPJ Quantum Technology, 6(1), 2.
49. Rosenberg, D., Kim, D., Das, R., Yost, D., Gustavsson, S., Hover, D., … & Oliver, W. D. (2017). 3D integrated superconducting qubits. npj Quantum Information, 3(1), 42.
50. McDermott, R., Vavilov, M. G., Plourde, B. L. T., Wilhelm, F. K., Liebfried, P. J., Simmonds, R. W., & Martinis, J. M. (2005). Quantum–classical interface for quantum information processing. Quantum Information Processing, 4(6), 465-492.

Figures

1. https://www.insight-ihbar.com/QC_hardware_en/F.html 
2. https://www.insight-ihbar.com/QC_hardware_en/F.html 
3. https://link.springer.com/article/10.1140/epjqt/s40507-025-00405-7 
4. https://www.nature.com/articles/s41467-019-10397-5 
5. https://www.researchgate.net/figure/Helium-3-reserve-and-disbursements-under-high-and-low-demand-scenarios-The-precipitous_fig1_284950269 
6. https://archive.org/details/gov.gpo.fdsys.CHRG-111hhrg57170/mode/2up 
7. https://scoop.market.us/quantum-computing-statistics/