Quantum computing exploits quantum mechanical phenomena (superposition, entanglement) to perform computations classical computers can't do efficiently — for specific problem types.
Hype is high; reality is more nuanced. This page covers what quantum computing actually is, what it can do, and what's realistic to expect.
0 or 1.
A superposition of 0 and 1. Mathematically: α|0⟩ + β|1⟩ where |α|² + |β|² = 1.
When measured: 0 with probability |α|², 1 with probability |β|².
A qubit is described by 2 complex numbers; n qubits by 2^n complex numbers. Quantum's exponential power.
Qubit is in multiple states simultaneously until measured.
Qubits become correlated such that measuring one determines the other.
Powerful for some algorithms; can't be created classically.
Quantum amplitudes can add or cancel. Algorithms manipulate amplitudes to cancel wrong answers and reinforce right ones.
Collapses superposition. Returns one classical outcome with probability determined by amplitudes.
Measurement is not "free observation" — it changes the state.
Can't copy an arbitrary unknown quantum state. Has implications for cryptography and error correction.
Quantum analog of classical logic gates. Reversible (must be — quantum mechanics is reversible).
A small set of universal gates can express any quantum computation (analog of classical NAND).
Factor large integers in polynomial time. Threatens RSA cryptography.
This is the algorithm that motivates quantum computing for cryptanalysis.
Search unstructured database in O(√n) instead of O(n). Quadratic speedup.
Affects symmetric cryptography (effectively halves key length).
Simulate quantum systems. Native task; classical computers struggle.
Likely the first practical application of quantum computers.
Solve linear systems exponentially faster (with caveats).
The caveats matter: input/output preparation can dominate.
Heuristic for combinatorial optimization. Mixed evidence on quantum speedup.
Find ground state of Hamiltonian. Used in chemistry simulation.
Various proposals; mostly research. No clear quantum advantage in current hardware.
Specific to algebraic problems. Doesn't apply to arbitrary tasks.
For unstructured search. Modest in practice.
Some problems show modest quantum advantages.
Most problems. Quantum doesn't speed up everything.
Quantum computers are physical systems with specific implementations:
IBM, Google, Rigetti. Cooled to millikelvin.
Currently the leading approach.
Quantinuum, IonQ. High-fidelity gates; slower.
PsiQuantum, Xanadu. Room temperature; harder to entangle.
QuEra, Atom Computing. Promising; growing.
Microsoft. Theoretically very stable; not yet demonstrated.
We are not at fault-tolerant quantum computing yet. We are at "noisy intermediate-scale quantum" (NISQ).
Demonstrating a task quantum computer does that classical can't simulate efficiently.
These tasks are usually not useful — chosen specifically to be classically hard.
The bar moves: classical algorithms improved (refuting some claims). Useful supremacy requires both quantum advantage AND useful work.
Qubits are noisy. To do reliable computation: error correction.
Logical qubit ≈ many physical qubits (typically 100-1000).
Required for fault-tolerant computation.
Currently demonstrated in research; not yet in operating systems.
Shor's algorithm breaks RSA and elliptic curve cryptography.
Symmetric cryptography (AES) is "only" weakened by Grover's. AES-256 → effectively AES-128 (still secure).
Cryptography secure against quantum computers. NIST standardizing:
Migration is happening; will take years.
Most experts: 10-30 years before cryptographically-relevant quantum computers.
But: "harvest now, decrypt later" is real concern. Adversaries collect encrypted data; decrypt when quantum computers exist.
Probably not for years. But long-lived secrets should consider post-quantum migration.
Optional. Tools exist (Qiskit, Cirq, Q#). Useful for specific roles (cryptography, quantum chemistry, research).
No. Quantum is for specific problem types. Classical computing remains primary.
Yes, on cloud quantum platforms (IBM, Amazon Braket, Azure Quantum). Limited but accessible.
Misleading. Quantum manipulates amplitudes; only specific algorithms produce useful results.
You can't just "search all solutions in parallel" generally.
Only for specific problems. Most computations don't benefit.
Not soon. Likely never for many applications.
Some understanding helps. Practical tools (Qiskit, Cirq) abstract much of it.
Most realistic near-term:
VQE and QAOA work this way.
In near term:
What probably won't be useful:
For most software engineers:
For interested engineers:
Quantum computing is real but not magic. Understanding what it can and can't do is more useful than overestimating either.