The most expensive part of a large quantum computer may in the end be the cabling. Finland has set out to run a machine with a thousand logical qubits by 2035. That would require hundreds of thousands of physical qubits — and millions of microwave lines to control them. Each costs around a thousand euros at today's prices, and each carries additional noise into a system that needs quiet.

At Aalto University in Finland, a component has now been built that could one day make that wiring harness redundant. Physicists led by Tuomas Uusnäkki and quantum technology professor Mikko Möttönen have operated the first cyclic heat engine inside a superconducting quantum circuit. Their results appeared on 13 July in Nature Communications.

The principle is two hundred years old; the scale is new. Heat engines — steam engines, combustion motors, power station turbines — convert a flow of heat into usable work. Physicists had already demonstrated this in the quantum regime with ultracold atoms and ions. But for superconducting circuits of all things, the basis of the quantum computers built by Google, IBM and Microsoft, the experimental realisation was missing.

At the heart of the Aalto setup is a transmon qubit in a cryostat near absolute zero. It is driven by a "quantum refrigerator," a superconducting circuit coupled to a resonator. Depending on the voltage applied, quasiparticles can tunnel through the circuit or not — and from precisely this the team obtains both states an engine needs: the qubit can be heated and cooled on demand. A single component thus plays the role of hot and cold reservoir at once, which makes the arrangement unusually lean.

Positive power, read off the magnetic field

With carefully timed pulses the team drove the qubit through an Otto cycle — the same process that runs in a car engine. The pulses alter the inductive energy of the circuit; the work performed can be read off the transmon's oscillation frequency and the system's magnetic field. The balance is what counts: less work flowed in than came out. The power reached more than 25 percent of the Otto efficiency.

The practical benefit lies not in propulsion but in control. If a fully autonomous machine could be built on this basis, it could read out qubits without microwave pulses having to be shuttled between millikelvin and room temperature. Much of the cabling — and much of its cost — would become dispensable. That becomes important precisely as qubit counts keep climbing, as practically useful applications demand.