What A Qubit Is, And What It Is Not

Every computer you have used stores information as bits. A quantum computer stores it as qubits, and most of the confusion about quantum computing starts with a loose description of what a qubit does. The popular line is that a qubit is “0 and 1 at the same time”, so a quantum computer must try every answer at once. That line is close enough to sound right and far enough off to mislead.

This article explains what a qubit actually holds, what happens when you read it and why the “every answer at once” story gives the wrong idea about what quantum computers can do. You need no physics or maths background, and the later articles in this series build on the ideas here.

Bits: The Familiar Starting Point

A classical bit has exactly two possible values, 0 or 1. A light switch is a fair picture: it is either off or on, and if you look at it you learn its state without changing it. Everything a laptop or a data centre does, from sending email to training an AI model, comes down to reading and flipping very large numbers of these switches.

Because a bit always has one definite value, copying it, checking it and storing it are simple. That simplicity is why classical computing is so reliable, and it is the baseline against which quantum computing should be judged.

Qubits: A Weighted Combination

A qubit can behave like a bit and hold a plain 0 or a plain 1. It can also be in a state that IBM describes as a weighted combination of both at once.¹ Physicists call this a superposition. NIST puts it even more simply: a qubit can be in state 0, state 1 or a mix of the two.²

The important word is “weighted”. A qubit in superposition carries two numbers called amplitudes, one attached to 0 and one attached to 1. Amplitudes are not ordinary probabilities. They can be positive or negative (and, in the full theory, complex numbers), which means they can add up or cancel out when qubits are combined. That ability to cancel is what quantum algorithms exploit, and it is the subject of the next article.

What Happens When You Measure

Reading a qubit is called measurement, and it is where the intuition from everyday life breaks down. This article describes the simplest and most common kind, which asks the qubit whether it is 0 or 1; physicists call it a standard basis measurement, and it is how a quantum computer normally reads out its answer.³ When a qubit in superposition is measured this way, the result is a single 0 or a single 1. The amplitudes decide the odds of each outcome, and the superposition does not survive the measurement.¹ Repeat the same measurement straight away, with nothing done to the qubit in between, and you get the answer you just got, not a fresh draw from the original mix.

So a qubit does not let you read out “both” values. You get one bit of output per qubit per run. To learn the odds themselves you have to prepare the same state many times and count the results, much as you would estimate whether a coin is biased by tossing it repeatedly.

Why “Tries Every Answer At Once” Is Wrong

Here is where the popular story goes astray. A group of qubits can be put into a superposition over a very large number of possible values. Scott Aaronson, a computer scientist who has written widely on the limits of quantum computing, points out that a thousand particles could in principle be described by around 10 to the power 300 such possibilities.⁴ That sounds like a machine that checks every answer in parallel.

The catch is measurement. Aaronson notes that when the qubits are read, you get just one of those possibilities, picked at random.⁴ If a quantum computer simply placed every candidate answer into superposition and measured, you would get a random candidate, which is no better than guessing. The size of the state does not turn into the same amount of readable output.

What makes a quantum algorithm useful is a careful choreography of the amplitudes. Because amplitudes can be negative as well as positive, they can cancel each other out. A good algorithm arranges its steps so that the amplitudes for wrong answers cancel and those for the right answer reinforce, so that measurement is likely to return something useful.⁴ This only works for problems with the right kind of hidden structure, which is why quantum computers are expected to be dramatically faster for some tasks and no faster for most.

Why Qubits Are Hard To Build

The same sensitivity that makes superposition useful also makes it fragile. Any unintended interaction with the outside world acts like a small, uncontrolled measurement and scrambles the amplitudes. NIST’s explainer, updated in May 2026, says the best machines today still make an error roughly once in every thousand operations, against roughly one error in a quintillion calculations for a classical computer.² Keeping qubits isolated, yet still controllable, is the central engineering problem of the field, and it is the reason error correction gets its own article in this series.

For security teams, the practical point is this: the risk quantum computing poses to today’s encryption does not come from qubits being magically parallel. It comes from one specific algorithm, published by Peter Shor in 1994, that uses interference to expose the hidden structure inside RSA and elliptic curve cryptography.⁵ The Quantum Threat section covers that risk in detail.