What is a logical qubit?
A logical qubit is a more reliable qubit built from many unreliable physical ones. Quantum error correction spreads one qubit's information across a group of physical qubits and keeps checking it for errors, so the group behaves like a single, much more dependable qubit. It only works if the physical qubits are already good enough, below an error rate called the threshold; above it, adding qubits makes things worse. Useful algorithms, including the ones that threaten encryption, are measured in logical qubits.
Physical versus logical
| Physical qubit | Logical qubit | |
|---|---|---|
| What it is | One piece of hardware: a superconducting circuit, a trapped ion, a neutral atom or a photon | A group of physical qubits working together under error correction |
| Reliability | Error rates vary by platform; on the best hardware, roughly one mistake every few hundred to few thousand operations | Far fewer errors, if the physical qubits are good enough |
| What headlines count | Usually this | The number that matters for useful algorithms |
How error correction works
You can't simply copy a qubit to make backups; quantum mechanics forbids it. Instead, error correction spreads the information across many qubits and repeatedly measures how neighbouring qubits relate to each other, without looking at the information itself. Those checks reveal where an error happened, so it can be fixed. The most studied design, the surface code, arranges qubits in a grid. The fragile qubits lesson explains it with a picture.
How many physical qubits make one logical qubit?
There's no single number. It depends on the error-correction code, how good the physical qubits are, how reliable the logical qubit needs to be, and the workload. Estimates are often in the hundreds of physical qubits per logical qubit. In 2024, Google reported that on its Willow chip, making the code larger made the logical error rate go down: a demonstration that below-threshold error correction works in practice, not proof that large cryptographic workloads are now practical.
To judge any quantum milestone, ask: physical or logical qubits, and at what error rate?
Why it matters for encryption
Published designs for breaking 2048-bit RSA use over a thousand logical qubits running for days. One 2025 estimate put the hardware at under a million noisy physical qubits, under specific assumptions; today's machines have hundreds to low thousands. A logical qubit count alone doesn't say whether a machine can run the attack: error rates, speed and running time matter too. The day a cryptographically relevant quantum computer becomes possible depends on progress in logical qubits, not on raw physical counts.
Common confusions
- "More qubits always means more power." Not if they're noisy. Quality matters as much as quantity.
- "A logical qubit is perfect." It's far better, not flawless. Bigger codes reduce errors further.
- "Companies count qubits the same way." They don't. A qubit in one hardware design isn't directly equivalent to one in another.
Sources and further reading
Tags show what kind of source each one is (how sources are used). Checked against these sources on 12 October 2026. Spotted an error? Email hello@plainquantum.com and it will be corrected, with a note.
- ExperimentQuantum error correction below the surface code thresholdGoogle Quantum AI and collaborators, Nature, 2024
- Peer-reviewed paperQuantum Computing in the NISQ era and beyondJohn Preskill, Quantum, 2018
- PreprintHow to factor 2048 bit RSA integers with less than a million noisy qubitsCraig Gidney, 2025