For quantum processors built around superconducting circuits, the refrigerator is part of the computing infrastructure. Helium-3 helps create the extreme cold these systems require.
Building a larger quantum computer means finding space for more than processors. The cooling system must also accommodate the connections that control the chips and carry their signals, while maintaining temperatures only a small fraction of a degree above absolute zero.
In August 2026, IBM reported connecting and cooling two cryogenic modules in a shared environment, reaching a final temperature below 15 millikelvin. The company described the work as a step towards a modular architecture for connecting quantum chips.
Fifteen millikelvin is 0.015 kelvin. Reaching that temperature, and maintaining useful cooling while equipment operates, requires an approach far beyond conventional refrigeration. Helium-3 is one of the materials that makes it possible.
How dilution refrigeration produces extreme cold
Dilution refrigerators exploit the behaviour of helium-3 mixed with helium-4. At sufficiently low temperatures, helium-3 can move from a concentrated phase into a more dilute phase. That process absorbs heat, providing cooling at the coldest stage of the refrigerator.
Researchers at the UK’s ISIS Neutron and Muon Source have used neutron imaging to examine helium movement inside a dilution refrigerator. Their work explains why the isotope’s behaviour matters to the performance of the complete system: the cooling depends on what happens within the mixture, not simply on placing a bottle of gas beside a processor.
The isotope is therefore part of an engineered circuit. Gas handling, heat exchange and the cold assembly have to work together to remove heat from the experiment or device.
The helium is circulated, not burned
Bluefors describes its dilution refrigerators as closed-loop systems using a helium-3 and helium-4 mixture. This makes an important difference to the supply discussion. A refrigerator requires an inventory of gas, but that inventory circulates through the cooling process rather than being consumed like a combustion fuel.
The manufacturer also publishes different helium-3 requirements for different systems. A single quantity per quantum computer would therefore be a poor basis for forecasting the whole market. The refrigerator design and configuration matter.
For an operator, retaining usable gas and supporting the equipment over its life are part of the supply problem. For a prospective supplier, the relevant opportunity includes new installations and inventory requirements, rather than an assumption that every hour of computation requires a fresh charge.
Scaling brings a cooling and connection challenge
IBM’s modular work illustrates another constraint. The company reported substantially more wiring space in each new vacuum enclosure than in its most widely used quantum systems. Its aim is to make room for more connections within and between modules.
That links cryogenic engineering directly to computing architecture. A system has to provide a workable environment for processors and their supporting connections, not merely achieve a low temperature when empty. Base temperature, cooling capacity and space for integration answer different questions about its usefulness.
The helium-3 discussion also needs to identify the type of quantum hardware involved. Superconducting circuits are the focus here; quantum computing includes other approaches with different operating requirements. Growth in the sector cannot be translated automatically into one universal helium-3 consumption rate.
Why supply agreements are drawing attention
In September 2025, Bluefors and Interlune announced a helium-3 purchasing agreement for up to 10,000 litres annually, with deliveries planned for 2028-2037. The announcement connects a proposed lunar source to a company already supplying cryogenic systems.
The agreement is a forward-looking supply arrangement. Its significance is that an equipment manufacturer is considering access to material over a longer planning period; the announcement does not demonstrate completed deliveries from the Moon.
As quantum hardware develops, the practical supply questions become more precise. How much inventory does a particular system need? What specification must it meet? Can existing material be retained or recovered? When would additional installations require delivery?
Those questions connect helium-3 supply to equipment that engineers can specify, and operators can plan for. They also give a more useful picture of the opportunity than treating every new qubit as a fixed amount of gas demand.
