The Quantum Investor's Notebook logo The Quantum Investor's Notebook
← Back

Friday, June 5, 2026

Qubit Count: The Scoreboard That Isn't

Qubit countThe raw number of physical qubits in a machine. A weak measure of power on its own, since quality and connectivity matter more. is the easiest number in quantum computing to understand. That is exactly what makes it dangerous.

The temptation of a single number

A 100-qubit machine sounds bigger than a 50-qubit one, and a 1,000-qubit machine more impressive still. The number is attractive because it looks like a scoreboard, one figure that lines every machine up and ranks them. But it measures only width. It says nothing about whether those qubitsThe basic unit of a quantum computer. Like a 'bit' in a normal computer, but instead of being only 0 or 1 it can be 0, 1, or a blend of both at once. can hold a calculation together long enough to finish it.1

A 50-qubit machine riddled with errors is, in practice, smaller than a clean 20-qubit one.

What the count leaves out is everything that decides whether the answer survives. Gate fidelityA score (for example 99.9%) for how accurately a quantum gate does what it is supposed to do. Higher means fewer errors. is how close each operation comes to the one you meant; as it slips, errors pile up until the output is indistinguishable from noise. ConnectivityWhich qubits in a machine can directly interact with each other. More connectivity makes more algorithms possible. sets which qubits can act on each other directly, and poor connectivity forces extra SWAP gatesAn operation that exchanges the states of two qubits, built from three CNOT gates. Used to shuttle information across a chip when the qubits it needs are not directly connected. that add still more error. And circuit depthThe number of sequential layers of operations a quantum computer can run before noise destroys the answer. The depth of a calculation, as opposed to its width., the number of operation layers a machine can run before the result decays, is the dimension a bare count ignores entirely.2

Two machines with the same headline. A wide but error-riddled machine fits only a small usable square of computation, while a narrower but cleaner machine fits a larger one. The usable square, not the width, is what counts.
Figure 1: Width is only one side of the square. A bigger but noisier machine can do less than a smaller, cleaner one, because the usable computation is limited by error, not by qubit count.

How you actually measure a machine

The honest metrics fold these together. Quantum volumeA single-number benchmark that folds together qubit count, gate fidelity, and connectivity by finding the largest square circuit a machine can run reliably. Higher is better. asks for the largest square circuit, as wide as it is deep, that a machine can run and still get the right answer more than two-thirds of the time. It is a single number, but one that already pays for weak fidelity and poor connectivityWhich qubits in a machine can directly interact with each other. More connectivity makes more algorithms possible., which is why it can be compared across very different hardware. When one group measured it, their best device of the day reached a quantum volume of 16, while others on the same scoreboard failed the test outright.2

Width times depth is about one over the error rate. Lower the errors and the usable machine grows in both directions at once.

Volumetric benchmarksA family of tests that measure a machine's width (how many qubits) and depth (how many operation layers) separately, instead of collapsing them into one number. go further, stretching width and depth separately so the trade-off is visible instead of hidden in one headline.3 They expose something a count never could: to reach the same quantum volume, an all-to-all machine can tolerate a two-qubit error rate of about 0.003, while a grid layout needs roughly half that.2 Other tools, like randomized benchmarkingA standard way to measure gate error by running random gate sequences and watching how fast the results decay., isolate the average gate error on its own, cleanly enough to compare across machines.4

Why it matters

The lesson for anyone keeping score is that progress is a move from quantity to quality. Adding qubitsThe basic unit of a quantum computer. Like a 'bit' in a normal computer, but instead of being only 0 or 1 it can be 0, 1, or a blend of both at once. does not bring a useful machine any closer if each new one is as noisy as the last; the road to error correctionTechniques that combine many shaky physical qubits into fewer reliable ones, so a long calculation stays correct. and fault toleranceThe milestone where a quantum computer can run long calculations correctly despite ongoing errors. It is the field's holy grail. runs through better gates and smarter compiling, not a bigger headline. The right question is never how many qubits a company has. It is what those qubits can actually do together.

Sources

  1. Preskill, J. “Quantum Computing in the NISQ Era and Beyond.” Quantum 2, 79 (2018). DOI: 10.22331/q-2018-08-06-79. Preprint: arXiv:1801.00862.
  2. Cross, A. W., Bishop, L. S., Sheldon, S., Nation, P. D. & Gambetta, J. M. “Validating Quantum Computers Using Randomized Model Circuits.” Physical Review A 100, 032328 (2019). DOI: 10.1103/PhysRevA.100.032328. Preprint: arXiv:1811.12926.
  3. Blume-Kohout, R. & Young, K. C. “A Volumetric Framework for Quantum Computer Benchmarks.” Quantum 4, 362 (2020). DOI: 10.22331/q-2020-11-15-362. Preprint: arXiv:1904.05546.
  4. Eisert, J., Hangleiter, D., Walk, N., Roth, I., Markham, D., Parekh, R., Chabaud, U. & Kashefi, E. “Quantum Certification and Benchmarking.” Nature Reviews Physics 2, 382–390 (2020). DOI: 10.1038/s42254-020-0186-4. Preprint: arXiv:1910.06343.