Quantum Computing vs Classical Computing: What Actually Changes
A quantum computer is not a faster version of the machine on your desk. It is a different kind of machine that is spectacularly good at a few problems and no help at all with most others.
Ask what a quantum computer is and you will often hear that it “tries every answer at once”. It is a tidy sentence, and it is wrong in a way that makes everything after it harder to understand.
This article compares the two kinds of computing properly: what the basic unit of information is, what a computation looks like, where the quantum advantage really comes from, and where there isn’t one.
The comparison at a glance
| Classical computing | Quantum computing | |
|---|---|---|
| Unit of information | The bit: 0 or 1 | The qubit: a state with two complex amplitudes |
| State of units | One of bit strings | A combination of all bit strings, with one amplitude each |
| Basic operations | Logic gates such as AND, OR and NOT | Reversible rotations of the state (unitary gates) |
| Reading the result | Look at the bits; nothing changes | Measure: you get one random outcome and the state is disturbed |
| Copying data | Free and exact | Impossible for an unknown state (no-cloning) |
| Errors | Extremely rare, fixed with simple redundancy | Common today; they need quantum error correction |
| Source of speed | More steps, done faster | Interference between amplitudes |
| Good at | Nearly everything we compute today | Factoring, simulating quantum systems, some search and optimization problems |
Bits and qubits
A bit is the simplest thing that can hold information: a switch that is off or on. Everything a classical computer does, from spreadsheets to video, is long strings of bits being rewritten according to rules.
A qubit is harder to summarize honestly. Its state is described by two numbers called amplitudes, one for 0 and one for 1. Amplitudes are not probabilities. They can be negative, or complex, and that single difference is what the whole field is built on. When you measure a qubit you get 0 or 1, with probabilities given by the squared sizes of the amplitudes, and the state you had is gone.
So a qubit is not “a bit that is 0 and 1 at once”. It is in one definite state. That state just isn’t one of the two you are able to read out.
What changes with many qubits
Describing bits takes numbers. Describing the state of qubits takes amplitudes, one for every possible bit string. For 50 qubits that is about a million billion numbers. For 300 it is more than the number of atoms in the observable universe.
This is where the “tries everything at once” story comes from, and where it goes wrong. Yes, a gate applied to qubits updates all amplitudes in a single step. But you can never read those amplitudes. A measurement returns one bit string. If all you did was spread a computation across every possible answer and then measure, you would get one answer at random, which you could have had by flipping coins.
Where the speedup really comes from
The useful part is interference. Because amplitudes can be negative, two computational paths that lead to the same outcome can cancel each other, so that outcome is never observed. Other paths reinforce.
A quantum algorithm is a carefully arranged sequence of gates that makes the paths to wrong answers cancel and the paths to right answers add up, so that when you finally measure, the outcome you want is likely. Arranging that cancellation is difficult, and it is only known to be possible for problems with the right kind of structure. That is why there are a handful of famous quantum algorithms rather than a quantum version of every program.
What a quantum computer is good at
- Factoring and discrete logarithms. Shor’s algorithm factors large numbers dramatically faster than any known classical method. Much of today’s public-key cryptography relies on factoring being hard, which is why this one algorithm gets so much attention.
- Simulating quantum systems. Molecules and materials obey quantum mechanics, and simulating them classically becomes exponentially expensive as they grow. A quantum computer represents them natively. Many researchers expect quantum simulation to be the first practical use.
- Unstructured search. Grover’s algorithm finds an item among possibilities in about steps. That is a real gain, but a modest, quadratic one.
- Some optimization and sampling problems. This is an active research area with fewer guarantees. Claims here deserve more scepticism than the three above.
What it is not good at
Most computing. A quantum computer will not make your browser faster, run your database, or render video. For the vast majority of tasks there is no known quantum algorithm that beats a good classical one, and classical hardware is cheaper, faster per operation, and vastly more reliable.
Nor is it a machine that solves every hard problem. The class of problems a quantum computer can solve efficiently is called BQP, and it is not believed to contain the notoriously hard NP-complete problems, such as finding the best route through many cities. For those, the known quantum gain over brute-force search is only quadratic.
Three things classical computers do that quantum ones can’t
- Copy. You copy files without thinking about it. An unknown quantum state cannot be copied, a result known as the no-cloning theorem. This is a law of physics, not an engineering gap, and it is also what makes quantum cryptography possible.
- Remember. Bits on a disk last for years. Qubits lose their state quickly, often in well under a second, through unwanted contact with their surroundings.
- Stay reliable without heroic effort. A classical chip performs billions of operations a second for years with essentially no logic errors. The best quantum gates today still fail far too often to run a long algorithm directly. The fix is quantum error correction, which combines many physical qubits into one dependable one, and it is the main engineering challenge of the field.
Will quantum computers replace classical computers?
No, and nobody building them expects that. The realistic picture is a quantum processor working beside classical ones, the way a graphics card does: a specialist called in for the few jobs it is good at, while a classical computer prepares the problem, controls the hardware and makes sense of the result. Even a quantum algorithm is usually wrapped in classical steps.
Common questions
Is a quantum computer faster than a supercomputer?
For a small number of specific problems it could in principle be far faster, to the point of doing in hours what would take a supercomputer longer than the age of the universe. For everything else it is slower. “Faster” only means something once you name the problem.
What is the difference between a bit and a qubit, in one sentence?
A bit is either 0 or 1; a qubit has an amplitude for 0 and an amplitude for 1, those amplitudes can interfere, and measuring it yields a single 0 or 1.
Do quantum computers exist today?
Yes. Machines with anywhere from tens to more than a thousand physical qubits exist in laboratories, and several can be used through the cloud. They are still too error-prone for the large algorithms that would beat classical computers on practical problems. Closing that gap is what the current race is about.
Where should I start if I want to understand this properly?
With a single qubit and something you can play with. Our free quantum computing course starts there, and our roadmap for learning quantum computing explains what to learn in which order.