How are quantum computers built? Qubit types compared

Superconducting qubits, trapped ions, neutral atoms, photons and spin qubits: what quantum computers are made of, who chose which path, and the pros and cons.

The short answer

A quantum computer is built from small physical systems with two states that can preserve quantum behaviour: qubits. There is no single “right” way. Superconducting circuits, trapped ions, neutral atoms, photons and semiconductor spins are the main approaches pursued by companies and labs. All of them are programmed the same way, with quantum circuits.

What makes a good qubit?

A qubit must keep its information for a long time (coherence time), gates must be fast with low error, qubits must be connectable and reliably measurable, and above all the platform must scale to thousands or even millions of qubits. Every hardware type excels at some of these criteria and lags at others.

Superconducting qubits

The approach of IBM, Google and Rigetti. The qubit is a tiny electrical circuit with a Josephson junction, cooled to just above absolute zero (about 10–20 millikelvin); the most common type is the transmon. Gates are very fast, on the order of nanoseconds, and chips can be made with semiconductor fabrication techniques.

Weaknesses: it needs a dilution refrigerator, coherence times are relatively short and each qubit usually interacts only with its neighbours. Google’s 2019 quantum supremacy experiment and its 2024 below-threshold error-correction result were achieved on this hardware.

Trapped ions

The approach of Quantinuum and IonQ. Individual atomic ions, levitated in a vacuum by electric fields, serve as qubits and are controlled with lasers. Because all ions are identical, the qubits are of very high quality, gate errors are among the lowest and every qubit can interact with every other.

Weaknesses: gates are much slower than on superconducting chips, and managing many ions in a single trap gets hard; complex trap architectures that shuttle ions are being developed to scale.

Neutral atoms

The approach of QuEra, Pasqal and Atom Computing. Uncharged atoms are held in two- or three-dimensional arrays by focused laser beams called optical tweezers and made to interact by exciting them to Rydberg states. Arrays of hundreds or even thousands of atoms, and the ability to move atoms around, are major advantages.

Weaknesses: gate speed and atom loss remain issues; the technology is maturing quickly. Quantum Store’s Pulser package shows how to program Pasqal’s neutral-atom devices.

Photons

The approach of Xanadu, PsiQuantum and Quandela. The qubit is stored in properties of individual particles of light (photons). Photons interact weakly with their environment, so they preserve information well; parts can operate near room temperature and connect naturally to fibre networks.

Weaknesses: making photons interact with each other is hard, and photon loss is the largest source of error. Quantum Store’s Perceval package shows how to simulate Quandela’s photonic circuits.

Spin qubits and topological qubits

Spin qubits: the spin of single electrons or nuclei in silicon is used as a qubit. Because they can be made in today’s chip fabs and are very small, companies such as Intel are exploring this path.

Topological qubits: the path Microsoft is following. The aim is to store information in “topological” states that are inherently robust against errors. Theoretically very attractive, but still at an early experimental stage and debated in the scientific community.

Which one will win?

Nobody knows yet. The real test for every approach is reaching large-scale logical qubits that run with error correction. The good news on the programming side: circuits written with tools such as Qiskit, Cirq, PennyLane and TKET can run on different hardware with small changes.

Frequently asked questions

Why are quantum computers cooled so much?

Superconducting qubits are very sensitive to thermal noise; to preserve their quantum states they must be cooled to a few hundredths of a degree above absolute zero. Ion and neutral-atom systems use vacuum and laser cooling instead.

Which quantum computer has the most qubits today?

The number of qubits alone is not a good measure; error rates and connectivity matter at least as much. Neutral-atom systems have reached arrays of thousands of atoms and superconducting chips more than a thousand qubits.

Can I simulate quantum computer hardware?

Yes. You can simulate the energy levels of superconducting transmon qubits with scqubits, neutral-atom arrays with Pulser and photonic circuits with Perceval on your own computer.

Concepts in this guide

Qubit · A qubit is the basic unit of information in a quantum computer: measured, it gives 0 or 1, but before measurement it can be in a superposition of both.
Decoherence · Decoherence is a qubit gradually losing its superposition and entanglement, its quantum properties, through unwanted interaction with its environment.
Quantum error correction · Quantum error correction spreads one logical qubit over many physical qubits so that errors caused by noise can be found and fixed without measuring the qubits directly.
Quantum supremacy · Quantum supremacy is the demonstration that a quantum computer performs a specific task that the most powerful classical computers cannot complete in reasonable time.
NISQ (noisy intermediate-scale quantum) · NISQ is the current era of quantum computing, named by John Preskill in 2018: devices with tens to a few hundred noisy qubits and no error correction.
Surface code · The surface code arranges qubits on a two-dimensional grid and uses only operations between neighbouring qubits; it is the most widely tested quantum error-correcting code today.

Full quantum computing glossary →