Skip to content
Tutorials Center
My learning

Topics

Lesson 4 of 39 ยท Getting Started

Quantum Hardware Landscape

Tour the physical qubit technologies (superconducting, trapped ion, neutral atom, photonic, spin), the DiVincenzo criteria, and the metrics that tell real devices apart.

A qubit is an abstraction: any physical system with two distinguishable states that can also exist in a controllable superposition. Many real systems qualify, and nobody yet knows which will win. This lesson gives you a map of the main technologies, a short history, and the handful of numbers you should check whenever someone announces a new quantum computer.

You do not need to remember the physics of each platform. What matters for a programmer is understanding what the hardware limits are, because they shape what circuits you can run.

A short history

  • 1981: Richard Feynman argues that simulating quantum physics efficiently needs a quantum machine.
  • 1985: David Deutsch describes a universal quantum computer.
  • 1994: Peter Shor publishes a factoring algorithm that would break common public-key cryptography on a large enough machine.
  • 1996: Lov Grover publishes his search algorithm.
  • 1995 to 1996: Quantum error-correcting codes show that noise can in principle be tamed.
  • 1998 to 2010s: First few-qubit demonstrations in NMR, trapped ions and superconducting circuits.
  • 2016 onward: Cloud access to small machines lets anyone run circuits.
  • 2019 onward: Experiments claim "quantum advantage" on carefully chosen sampling tasks, followed by improved classical simulations and a debate that continues.
  • 2020s: The field's focus moves from raw qubit counts to error rates and the first demonstrations of error correction improving as more qubits are added.

The main qubit technologies

flowchart TB
    Q["Physical qubit"]
    Q --> S["Superconducting circuits<br/>(transmons)"]
    Q --> I["Trapped ions"]
    Q --> N["Neutral atoms<br/>(optical tweezers)"]
    Q --> P["Photonics"]
    Q --> SP["Spin qubits<br/>(silicon, NV centers)"]
    Q --> T["Topological<br/>(experimental)"]

Several unrelated physical systems are being developed in parallel.

Superconducting qubits are tiny electrical circuits cooled to a few hundredths of a degree above absolute zero. Their gates are very fast (tens of nanoseconds) and they are made with chip-fabrication techniques, but each qubit has a short lifetime and connects only to nearby neighbors. This is the technology behind many of the cloud machines of large companies.

Trapped ions use individual charged atoms held by electric fields in vacuum, with lasers as gates. Their qubits last long and any ion can interact with any other in the same trap, and their gates are very accurate, but gates are slower (microseconds and up) and scaling a single trap is hard.

Neutral atoms are held by tightly focused laser beams ("optical tweezers") in large, reconfigurable arrays. They scale to hundreds or thousands of atoms and arrangements can be changed between operations.

Photonic machines encode information in light. Photons barely interact with their environment, and operate at room temperature in parts, but making two photons interact on demand is the central difficulty.

Spin qubits store information in the spin of an electron or nucleus in silicon or in defects such as nitrogen-vacancy centers in diamond. They promise compatibility with the semiconductor industry.

Topological qubits would store information in global properties of exotic matter, making them naturally robust. As of today they remain experimental.

Separately, quantum annealers are a special-purpose design that solves optimization problems by slowly evolving a system into a low-energy state. They are not general gate-based computers, and this course focuses on gate-model machines.

The DiVincenzo criteria

In 2000 David DiVincenzo listed what any candidate technology needs, and the list is still the best checklist:

  1. A scalable collection of well-characterized qubits.
  2. The ability to initialize them to a simple state such as all zeros.
  3. Coherence times much longer than gate times.
  4. A universal set of gates (see Universal Gate Sets and Circuit Identities).
  5. A qubit-specific measurement capability.

Metrics worth checking

  • Number of qubits. The headline figure, and the least informative on its own.
  • Gate fidelity. The probability a gate does what it should. 99.9 percent sounds excellent, yet errors compound over many gates.
  • Coherence time (T1, T2). How long a qubit keeps its state, compared to the gate time.
  • Connectivity. Which qubit pairs can interact directly. Poor connectivity forces extra swap gates.
  • Readout error. How often a measurement returns the wrong bit.
  • Circuit depth achievable. How many layers of gates fit before the answer is noise.

The first stand-in below shows why fidelity matters more than qubit count. With a per-gate success probability f, a circuit with g gates succeeds fully only about f^g of the time (a rough model that assumes independent errors).

Python How per-gate fidelity limits circuit size

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The second ratio that matters is how many gates fit inside the qubit's lifetime. A technology with fast gates and a short lifetime can be as limited as one with slow gates and a long lifetime.

Python Gates per coherence time (illustrative numbers)

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The numbers are deliberately rounded and illustrative. The lesson is the method: always compare coherence time to gate time, and look at the reported error rates, not only qubit counts.

Whatever the technology, today's machines are called NISQ devices (noisy intermediate-scale quantum). Chapter 6 explains noise and how error correction aims to move beyond NISQ.

What this means for you as a programmer

  • Circuits on real hardware must be short; expect to transpile (rewrite) them for the device's native gates and connectivity.
  • Results are statistical: you run a circuit many shots and look at counts.
  • Simulators (like the ones you will build in this course) are noise-free unless you add noise yourself.
  • Everything you learn about gates, entanglement and algorithms is hardware independent. Only the cost model differs.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Which of these is a quantum computing hardware platform?

    Trapped ions are one of the leading qubit technologies, together with superconducting circuits and neutral atoms.
  2. 2. Why is qubit count alone a poor measure of a machine?

    A machine with many noisy, poorly connected qubits may run shallower useful circuits than a smaller, cleaner one.
  3. 3. A gate with fidelity 0.999 is applied 1000 times. Roughly what is the chance of no error (independent errors)?

    0.999^1000 is roughly e^-1, about 0.37.
  4. 4. Which DiVincenzo criterion says qubits must keep their state for much longer than a gate takes?

    If gates are slower than the decoherence, information is lost before the computation finishes.

Summary

  • Many physical systems can serve as qubits; no clear winner exists yet.
  • Superconducting circuits are fast, trapped ions are accurate and well connected, neutral atoms scale in large arrays.
  • DiVincenzo's criteria list what any technology needs.
  • Fidelity, coherence-to-gate-time ratio, connectivity and readout error matter more than raw qubit count.
  • Today's devices are noisy (NISQ), so circuits must be short and results statistical.

Next: The Qubit and Its State Vector