In short: A quantum computer stores information in qubits that can sit in a superposition of 0 and 1, then uses interference to amplify the amplitudes of correct answers and cancel wrong ones, which gives it a real advantage on a narrow set of problems such as factoring and simulating molecules, not on everyday computing.
Why interviewers ask this question
"What is quantum computing?" is a favourite opener because a weak answer is easy to spot. Candidates who say "it tries every answer at once" have repeated the most common myth. A strong answer names three ingredients (qubits, superposition, interference), explains that measurement destroys most of the information, and points to the specific problems where the maths gives a speed-up. This article walks through that answer the way you would deliver it on a whiteboard. For the head-to-head view, see our companion piece classical vs quantum computing.
From bits to qubits
A classical bit is 0 or 1, held as a voltage across a transistor (roughly 0 V or around 0.7 V in a modern core supply). A qubit is a two-level quantum system, for example the two lowest energy states of a superconducting circuit or the spin of a trapped ion. Its state is written as α|0⟩ + β|1⟩, where α and β are complex numbers called amplitudes and |α|² + |β|² = 1.
The key detail is that amplitudes are complex, so they carry a phase as well as a size. Phase is invisible if you measure straight away, but it decides what happens when paths recombine. That is the whole source of quantum power.
Superposition is not "trying everything"
With N qubits the machine tracks 2^N amplitudes at once. For 50 qubits that is about 1.1 × 10^15 numbers; storing them in a classical simulator at 16 bytes each needs roughly 18 petabytes of memory. That exponential bookkeeping is why classical simulation of even modest quantum circuits runs out of room, and why people say quantum machines "hold" huge state.
But when you measure, you get one N-bit string, chosen randomly with probability equal to the squared magnitude of its amplitude (the Born rule). You cannot read out all 2^N values. So superposition alone gives nothing; an algorithm has to shape the amplitudes before you look.
Key fact: A quantum computer does not return 2^N answers. It returns one, and the algorithm is designed so that the right one is overwhelmingly likely.
Interference does the real work
Amplitudes add like waves. Two computational paths leading to the same outcome can reinforce (same phase) or cancel (opposite phase). A quantum algorithm is a choreography of gates that makes paths to wrong answers cancel and paths to right answers add up.
The smallest demonstration uses the Hadamard gate H. Applied to |0⟩ it gives an equal superposition, so a measurement is a fair coin flip. Apply H a second time and the two paths into |1⟩ carry opposite signs and cancel, returning |0⟩ with certainty. A classical coin flipped twice would stay random; the qubit un-randomises itself. The animation below shows probabilities at each stage.
Problems where the maths helps
Quantum advantage is proven or strongly suspected only for specific problem structures:
- Factoring (Shor, 1994): finds the period of a modular function using the quantum Fourier transform. It runs in polynomial time, while the best classical method (the number field sieve) is sub-exponential. This threatens RSA and elliptic-curve key exchange.
- Unstructured search (Grover, 1996): a quadratic speed-up, about √N steps instead of N. Useful, but it only halves the effective bit-strength of symmetric keys, which is why AES-256 remains considered safe.
- Simulation of quantum systems (Feynman's original motivation): molecules, catalysts and materials whose electron correlations explode classically.
Key fact: Breaking RSA-2048 is estimated to need on the order of a million or more physical qubits once error correction is counted, versus the hundreds of noisy qubits available today.
Why it is still hard: noise and error correction
| Item | Typical figure | Consequence |
|---|---|---|
| Two-qubit gate error | about 0.1% to 1% | Circuits of thousands of gates fail without correction |
| Coherence time (superconducting) | tens to hundreds of microseconds | Algorithm must finish before the state decays |
| Operating temperature | about 15 mK | Dilution refrigerators, heavy cabling |
| Physical per logical qubit (surface code) | hundreds to ~1,000 | Useful machines need millions of physical qubits |
Because qubits cannot be copied (no-cloning theorem), error correction spreads one logical qubit across many physical ones and keeps measuring syndromes. That overhead, not the qubit count on a press release, decides when Shor becomes practical.
The silicon angle for chip engineers
Quantum processors are not built from standard logic, but they live next to conventional chips. The control electronics that generate microwave pulses and read out qubits are increasingly cryogenic CMOS operating at 4 K, designed in tools such as Cadence Virtuoso and Spectre and Synopsys custom flows, with device models re-characterised at low temperature. Spin-qubit work even reuses FinFET-like silicon processes. If you want the fabrication and design-flow side of this field, our sister training brand ChipXpert VLSI training covers the digital and analog skills this hardware depends on. For how these ideas connect to cryptography, read post-quantum cryptography explained.
Ready to go beyond the interview answer? Explore the hands-on electronics, VLSI and emerging-technology tracks in our CourseTron courses and build the fundamentals employers test for.
Frequently asked questions
Can a quantum computer replace my laptop?
No. It is an accelerator for specific problems. For web browsing, databases or most machine learning workloads, classical hardware is faster and far cheaper.
Does a quantum computer try all answers at once?
No. It holds amplitudes for all answers, but a measurement returns one outcome. Useful algorithms use interference so the correct outcome is the likely one.
Can quantum computers break encryption today?
Not yet. Breaking RSA-2048 with Shor's algorithm needs far more error-corrected qubits than exist now. Standards bodies are already migrating to post-quantum algorithms to prepare.
What is the difference between a qubit and a bit?
A bit is definitely 0 or 1. A qubit is described by two complex amplitudes, so it can be in a superposition, and its phase lets different computational paths interfere.