What is quantum computing? Qubits in plain words

Quantum computing is a way of computing that uses the rules of quantum physics instead of ordinary on-off switches. Its basic unit, the qubit, can hold a blend of 0 and 1, and many qubits can be linked so that their states depend on each other. For a few kinds of problems, such as simulating molecules or factoring huge numbers, that could make a quantum computer vastly faster. For most everyday tasks, it would be no faster at all.

In short

• A qubit holds a weighted blend of 0 and 1 until it is measured, and linked qubits describe far more possibilities than the same number of bits.
• Quantum computers are not faster at everything; they win only on problems with a structure their algorithms can exploit.
• The hard part is errors: qubits are fragile, and correcting their mistakes takes many physical qubits for each reliable one.

What is a qubit?

An ordinary computer stores information in bits, each of them 0 or 1, held by transistors that are either on or off, as ordinary computers do everywhere. A qubit, or quantum bit, is a physical system with two basic states that can also be in a superposition of both: a combination with a weight attached to 0 and a weight attached to 1. The weights, called amplitudes, can be positive, negative or more complicated numbers, and they decide the odds of what you'll see.

The catch is measurement. When you read a qubit, you never see a blend: you get a plain 0 or a plain 1, with odds set by the amplitudes, and the superposition is gone. So a qubit isn't a secret store of two answers you can read out. Superposition comes straight from quantum mechanics, which describes particles such as electrons and photons this way. Real qubits are built from tiny superconducting circuits cooled to near absolute zero, single charged atoms held in place by electric fields, particles of light and other systems.

Entanglement is the second ingredient. Two or more qubits can be linked so that their states can only be described together. Measure one of a pair of entangled qubits and you know something about the other, however far apart they are. Entanglement is what lets a group of qubits represent patterns that no list of separate bits can.

Why do qubits add up so fast?

Here is the arithmetic that makes quantum computing interesting. Describing the state of n qubits takes 2 to the power n amplitudes, one for every possible pattern of 0s and 1s. Each extra qubit doubles the count.

Qubits	Amplitudes needed to describe them	Memory to store them on an ordinary computer (8 bytes each)
10	1,024	About 8 kilobytes
30	About 1.1 billion	About 8.6 gigabytes
53	About 9 million billion	About 72 million gigabytes
300	About 2 followed by 90 zeros	More than there are atoms in the observable universe

That is why a 53-qubit chip, Google's Sycamore, made headlines: simulating it means juggling about 9 quadrillion numbers. It is also why simulating nature is the most natural job for a quantum computer, an idea the physicist Richard Feynman put forward in a 1981 talk: molecules and materials obey quantum rules, and describing them on ordinary computers grows exponentially harder as they get bigger.

Doesn't a quantum computer just try every answer at once?

No, and this is the most common misconception. If it did, measuring would give one random answer out of all of them, which is useless. What quantum algorithms actually do is arrange interference: amplitudes behave like waves, so they can add up or cancel out. A good algorithm makes the paths leading to wrong answers cancel and the paths leading to right answers reinforce, so that when you measure, the right answer is likely.

Only some problems have the hidden structure that allows this. In 1994 the mathematician Peter Shor found an algorithm that factors large numbers, splitting them into the primes that multiply to make them, exponentially faster than any known method on ordinary computers. In 1996 Lov Grover found one that searches an unsorted list in about the square root of the usual number of steps: about 1,000 checks instead of up to 1,000,000. That speed-up is real but modest, and for many tasks, such as running a spreadsheet, a quantum computer would offer nothing.

Good fit for quantum computers	Poor fit
Simulating molecules and materials, for chemistry and drug design	Email, browsing, video, office work
Factoring large numbers (Shor's algorithm)	Tasks that need reading in huge amounts of data
Some optimization and search problems, with modest gains	Anything ordinary computers already do quickly

Why is error correction the hard part?

Qubits are extraordinarily fragile. Heat, stray electromagnetic fields or a passing vibration can nudge their amplitudes, a process called decoherence, and even good qubits today get an operation that links two of them wrong roughly once in every thousand or so tries. Useful algorithms such as Shor's need billions of operations without a mistake.

The answer is quantum error correction: spread one reliable "logical" qubit across many physical qubits, and check them constantly for signs of error without measuring the information itself. It works only if the physical qubits are already good enough, below a threshold; otherwise adding more qubits adds more errors than it fixes. Google reported in December 2024 that its 105-qubit Willow chip had crossed that line, halving the logical error rate each time it grew its grid of qubits from 3 by 3 to 5 by 5 to 7 by 7.

The milestones so far, each disputed in its own way:

When	What was claimed
1981	Richard Feynman proposes computers built on quantum rules, to simulate nature.
1994	Peter Shor publishes his factoring algorithm.
October 2019	Google says its 53-qubit Sycamore did a sampling task in about 200 seconds that would take a top supercomputer about 10,000 years. IBM researchers replied that a classical system could do it in 2.5 days.
August 2024	The US National Institute of Standards and Technology publishes its first encryption standards designed to resist quantum attack.
December 2024	Google reports error correction below threshold on Willow.

The tasks behind these milestones were chosen to be hard for ordinary computers, not to be useful. Whether any quantum computer has yet done useful work that ordinary computers can't is itself argued: several such claims have been followed by better classical methods that narrowed or closed the gap.

Will quantum computers break encryption?

Much of the internet's security rests on public key cryptography, which assumes that some math problems, such as factoring the product of two large primes, are too hard to solve. Shor's algorithm would break that assumption on a large enough, error-corrected quantum computer, exposing the encrypted traffic that crosses the internet every second. Hence the new standards: in August 2024, NIST released three post-quantum standards, FIPS 203, 204 and 205, built on math problems that no known quantum algorithm solves fast, and urged organizations to start switching as soon as possible.

Still argued

When, or whether, a quantum computer will break today's encryption is genuinely disputed. NIST notes that some experts expect such a machine within a decade. Estimates of what it would take keep falling: in May 2025, Google researcher Craig Gidney estimated that 2048-bit RSA keys could be factored in under a week with fewer than a million noisy qubits, down from 20 million in his 2019 estimate with Martin Ekerå. But the machines built by 2026 are far smaller than a million qubits, and most of their qubits are too noisy to work together reliably; skeptics argue that keeping millions stable may meet engineering limits no one has yet solved. What is agreed is the risk of "harvest now, decrypt later": data recorded today could be read once such a machine exists, which is why the switch to new standards has already begun.

Quantum computing is sometimes paired with artificial intelligence in headlines, but there is as yet little evidence that quantum computers will speed up the kind of AI in use today, which depends on processing vast amounts of data, a poor fit for them.

Check yourself

• Without looking back: why does adding one qubit double what it takes to describe a quantum computer's state?
• Why is "it tries every answer at once" wrong, and what does a quantum algorithm do instead?
• What does it mean for error correction to work "below threshold"?

Sources

• Quantum supremacy using a programmable superconducting processor — Arute et al., Nature, 2019
• On "Quantum Supremacy" — IBM Quantum, 2019
• Meet Willow, our state-of-the-art quantum chip — Google, 2024
• NIST releases first 3 finalized post-quantum encryption standards — National Institute of Standards and Technology, 2024
• Polynomial-time algorithms for prime factorization and discrete logarithms on a quantum computer — Peter W. Shor, 1995
• How to factor 2048 bit RSA integers with less than a million noisy qubits — Craig Gidney, 2025