Quantum Computing Explained
A quantum computer is not simply a faster classical computer. It is a fundamentally different kind of machine that exploits three quantum phenomena — superposition, entanglement, and interference — to solve specific problems that are intractable for any classical machine, regardless of its size. No formulas required.
Bit vs Qubit
Classical Bit
- A switch: either ON (1) or OFF (0)
- Exactly one state at a time
- 8 bits = one of 2⁸ = 256 possible values; holds exactly one
- Copy freely (no restrictions)
- Operations are deterministic
Qubit
- A quantum system with two distinguishable states (|0⟩ and |1⟩)
- Can be in a superposition — a blend of both
- 8 qubits = superposition over all 256 values simultaneously
- Cannot be copied (no-cloning theorem)
- Operations are reversible quantum gates
A qubit is physically implemented as anything with two quantum states: the spin of an electron (up/down), the polarisation of a photon (horizontal/vertical), the energy level of a superconducting circuit, or a trapped ion. The specific physical implementation defines which company's hardware you are looking at.
Superposition: Both At Once
The most common — and most misunderstood — quantum concept. A qubit in superposition is not secretly either 0 or 1 (and you just don't know which). It is genuinely in a state that is a combination of both, describable by two numbers ("amplitudes") that encode the probability of measuring each outcome.
(always H or T)
(H and T until measured)
A useful (but imperfect) classical analogy: imagine a coin spinning in the air. While spinning it is neither heads nor tails. When it lands (is "measured"), it collapses to one or the other. Unlike a real coin, quantum amplitudes are not just probabilities — they can be negative or complex, enabling interference.
Entanglement: Correlated Fates
Two qubits can be prepared in an entangled state: a joint superposition where the measured values of the two qubits are correlated — no matter how far apart the qubits are when measured.
The canonical example is the Bell state: two qubits prepared so that measuring the first always gives the same result as measuring the second (both 0, or both 1 — randomly chosen, but always matching). Einstein called this "spooky action at a distance."
Entanglement is not classical correlation (like matching gloves in separate boxes). The correlation exists before measurement — the qubits are not secretly predetermined to the same value; they are genuinely in a joint superposition. Bell's theorem (1964) and experimental tests confirm this.
In quantum computing, entanglement lets qubits that are never directly coupled influence each other, creating massive correlations across the entire quantum register that a classical computer would need exponential memory to represent.
Interference: Amplifying the Right Answer
Interference is the most important concept for understanding why quantum computers can solve certain problems. Quantum amplitudes behave like waves: they can add (constructive interference) or cancel (destructive interference).
A quantum algorithm is engineered so that wrong answers interfere destructively (their amplitudes cancel out) while right answers interfere constructively (their amplitudes add up). When finally measured, the register almost always collapses to the correct answer.
This is analogous to noise-cancelling headphones: the algorithm introduces carefully tuned "anti-noise" to mathematically cancel every incorrect computational path, leaving only the paths that lead to the solution.
Measurement: Collapse
Here is the catch: when you measure a qubit, its superposition collapses irreversibly to a definite classical value — 0 or 1 — according to its amplitude probabilities. You get one sample, not the full distribution.
This is why simply "looking at all the answers simultaneously" is not an algorithm. The art of quantum computing is engineering the interference pattern so that, by the time you measure, the superposition has evolved to concentrate almost all probability on the correct answer.
Algorithms often need to be run many times and the results aggregated — or they are designed to guarantee the correct answer with high probability in a single shot.
The Real Power: Exponential State Space
A register of n classical bits can hold one of 2ⁿ possible values. A register of n qubits can be in a superposition of all 2ⁿ values simultaneously. To completely describe this state on a classical computer, you would need 2ⁿ complex numbers — one amplitude for each basis state.
This exponential state space is why quantum computers cannot simply be simulated on classical machines beyond ~50 qubits — and why they have the potential for exponential speedup on specific problems.
What Quantum Computers Are Actually Good At
Quantum computers are not universally faster. They offer provable or strong expected speedups on specific problem structures:
| Problem | Classical best | Quantum speedup | Impact |
|---|---|---|---|
| Factoring large integers | Sub-exponential (billions of years for 2048-bit) | Polynomial (Shor's algorithm) | Breaks RSA/ECDH encryption |
| Unstructured search | O(N) | O(√N) (Grover's algorithm) | 2× more QBits to maintain classical security |
| Simulating quantum chemistry | Exponential — exact simulation impossible | Polynomial (VQE, QPE) | Drug discovery, materials, catalysts |
| Linear systems (HHL algorithm) | O(N) classical | O(log N) with caveats | Potential ML speedup (contested) |
| General optimisation | Heuristic (NP-hard) | Uncertain (QAOA) | Logistics, finance (unproven) |
| Sorting / general computation | Fast classical algorithms | No quantum advantage known | Classical computers stay better here |
Current Hardware
All current quantum computers are NISQ devices — Noisy Intermediate-Scale Quantum. "Noisy" because qubits have decoherence times of microseconds to milliseconds, meaning they lose their quantum state through environmental interactions before computations finish. "Intermediate scale" because today's machines have 50–1,000+ physical qubits but cannot yet perform the error correction needed for reliable long computations.
Leading hardware approaches:
- Superconducting qubits (IBM, Google): Electrical circuits cooled to ~0.015 K (colder than outer space). Gate speeds: ~10–100 ns; coherence: ~100 μs. IBM has >1000-qubit chips; Google demonstrated quantum supremacy (2019) with 53 qubits on a random circuit sampling task.
- Trapped ions (IonQ, Quantinuum): Individual atomic ions held by electromagnetic traps. Slower gates (~100 μs) but longer coherence (seconds) and better connectivity. Quantinuum H2 (2024) reached 56 qubits with full qubit-to-qubit connectivity.
- Photonic qubits (PsiQuantum, Xanadu): Photons as qubits. Room temperature possible; hard to make qubits interact deterministically.
- Neutral atoms (Atom Computing, QuEra): Arrays of up to 1000+ neutral atoms in optical tweezers. High connectivity; recent demonstrations of error correction.
Timeline and Realistic Expectations
Today (2026): NISQ devices can run short demonstrations of quantum advantage on contrived tasks. No practical quantum advantage over classical computers on real-world problems has been demonstrated yet.
Near term (2027–2032): Early fault-tolerant machines with a few thousand logical qubits (encoded in ~millions of physical qubits via error correction). First genuine quantum advantage in quantum chemistry likely in this window.
Cryptographically-relevant quantum computers: Breaking 2048-bit RSA is estimated to require ~4,000 logical (≈ 4 million physical) qubits doing ~10 billion gate operations. Current consensus: 10–20+ years away. This is why NIST published post-quantum cryptography standards in 2024.
Quantum computing is not hype — but it is also not "just around the corner" for most applications. The engineering challenge of building large-scale, fault-tolerant quantum computers is immense. The potential payoff, especially in drug discovery and materials science, is real and worth the investment.