Skip to content
Tutorials Center
My learning

Topics

Lesson 35 of 39 ยท Applications and Outlook

The Outlook for Quantum Computing: Advantage, Roadmaps and Hype

Where is quantum computing heading? Learn what quantum advantage means, how to estimate resources, which applications are likely, and how to spot hype.

After eight chapters you can read quantum circuits, run algorithms and understand why noise is the central obstacle. This lesson steps back and asks the practical questions: What can quantum computers realistically do? When? How do you judge a headline? You will also use two small calculators that turn vague claims into numbers.

Hardware and records change quickly. This lesson focuses on concepts that stay valid, and quotes numbers as rough orders of magnitude. For vendor details, check the current roadmaps; the hardware landscape lesson explains the qubit types.

What "quantum advantage" means

A quantum advantage (also called quantum supremacy in early papers) is a task where a quantum computer beats the best classical method in some resource, usually time. Three levels are useful:

  1. Advantage on a contrived task. Random circuit sampling and boson sampling were designed to be easy for quantum hardware and hard to simulate. Experiments since 2019 have claimed advantage here, followed by classical simulations that narrowed or closed the gap. The tasks have no known practical use, but they test that the machine does something classically hard.
  2. Advantage on a useful scientific task, for example simulating a quantum material or dynamics beyond classical reach. Recent experiments are approaching this, and they are carefully compared against improving tensor-network and other classical methods.
  3. Commercial advantage: solving a problem someone will pay for better or cheaper than any classical method. This has not been convincingly demonstrated.

A fair claim states the task, the classical baseline, and the cost (time, energy, money) on both sides.

Fault tolerance is the dividing line

Today's machines are NISQ. Algorithms with a known exponential speed-up (factoring, long chemistry simulations via phase estimation) need billions of operations, so they need logical qubits from error correction. Roadmaps from major companies aim for machines with hundreds of logical qubits around the end of the 2020s and thousands later, with much uncertainty. Demonstrations of error correction below threshold (logical error falling as code distance grows) are the critical milestone, and they have started to appear.

A rough overhead: one logical qubit at useful error rates takes hundreds to thousands of physical qubits. The calculator below turns that into totals.

Python How many physical qubits would an algorithm need?

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 logical-qubit numbers are rounded, illustrative figures; what matters is the scale. Better algorithms, better codes (quantum LDPC codes) and lower physical error rates keep pushing these totals down, so estimates have fallen over time. That trend is why security planners do not wait for a working machine before migrating to post-quantum cryptography (see Quantum Cryptography).

Why classical simulation is a moving target

A state vector of n qubits needs 2^n complex numbers, 16 bytes each in double precision. That is why laptops stop around 30 qubits, and supercomputers at about 45 to 50. But many circuits are not worst case: tensor networks, stabilizer simulators and Pauli-propagation methods can handle far larger systems if entanglement or noise is limited. Every advantage claim must survive these methods.

Python Memory needed to simulate n qubits exactly

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 table explains why quantum hardware with a few hundred good, well-connected qubits can be hard to simulate exactly, and also why exact simulation is not the only classical option.

Likely quantum winners

Chemistry Catalysts and batteries
Materials Superconductors, magnets
Cryptanalysis Shor against RSA and ECC
Quantum sensing Clocks and field sensors

All rely on quantum structure in the problem itself.

1 / 2

Applications, ranked by realism

  1. Simulating quantum systems (chemistry, materials, high-energy physics). The strongest case, with provable speed-ups once error-corrected machines exist.
  2. Cryptanalysis. A clear, well-understood speed-up from Shor's algorithm. A reason to adopt post-quantum cryptography today.
  3. Quantum sensing and metrology. Already commercial in some forms (atomic clocks, magnetometers). Not "computing", but a payoff of the same control technology.
  4. Optimization and finance. Possible polynomial speed-ups (Grover-type, amplitude estimation for Monte Carlo pricing); constant factors and data loading may erase them. Needs fault tolerance.
  5. Machine learning. Plausible for quantum data. For classical data, no convincing advantage; see quantum machine learning.
  6. Linear algebra and differential equations. HHL-type algorithms promise exponential gains with heavy caveats (input, output, conditioning).

How to read a quantum headline

Ask these questions:

  • What exactly was computed? A useful result, or a benchmark made for quantum hardware?
  • What was the classical baseline? The best known algorithm, or a naive one? Has anyone since simulated it?
  • What is the error? Was the output verified? How many shots, and what mitigation?
  • Logical or physical qubits? "1,000 qubits" in a press release means physical qubits, with error rates that limit usable circuit depth. Look at two-qubit gate fidelity, connectivity and coherence, not only the count.
  • What scales? One result at 100 qubits does not mean 1,000,000.
  • Who benefits from the claim? Funding and marketing influence wording on every side, including skeptics.

Useful metrics: two-qubit gate error, coherence times, connectivity, cycle time, quantum volume (a single-number benchmark of width and depth), and logical error rate per cycle once error correction appears.

Beyond gate-model computing

  • Quantum annealing (D-Wave) solves Ising/QUBO problems on thousands of qubits; whether it beats classical solvers is debated, though studies of quantum dynamics on annealers show promise.
  • Analog and neutral-atom simulators emulate specific Hamiltonians directly, useful for materials science.
  • Photonic computers use light and promise room-temperature operation and networking; error rates and loss are challenges.
  • Quantum networks and the quantum internet will link devices with entanglement distribution and repeaters, supporting distributed quantum computing and secure communication.
  • Topological qubits, still experimental, aim for built-in protection that would cut error-correction overhead.

Careers and learning paths

  • Software: SDKs, compilers, error-mitigation and decoder software, hybrid workflows (Qiskit, PennyLane, Cirq).
  • Theory and algorithms: complexity, error-correcting codes, resource estimation, chemistry mappings.
  • Hardware: superconducting circuits, ion traps, neutral atoms, photonics, cryogenics, control electronics.
  • Applications: chemistry, finance, logistics, security; many roles are "quantum-ready" classical engineers who prepare algorithms and data pipelines.

A solid base of linear algebra, probability, Python and some physics or computer science is enough to start. Practicing small simulations, as in the projects that follow, builds the intuition that matters most.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Why do algorithms like Shor's need error correction?

    With gate errors of 10^-3 or so, circuits longer than a few thousand gates fail; logical qubits are needed.
  2. 2. Which is the most realistic early application of quantum computers?

    Simulating quantum physics has the strongest theoretical case for advantage.
  3. 3. A press release says '1,000 qubits'. What should you check first?

    Qubit count alone says little; error rates, connectivity and logical qubits matter.
  4. 4. Why can't classical simulation be dismissed when evaluating an advantage claim?

    Many claims were later matched or narrowed by better classical algorithms.

Summary

  • Quantum advantage claims need a task, a baseline and a cost; contrived-task advantages are real but not yet commercial.
  • Algorithms with exponential speed-ups need fault-tolerant machines with thousands of logical and millions of physical qubits, though estimates keep falling.
  • Chemistry, materials and cryptanalysis are the most credible applications; machine learning and optimization remain open.
  • Evaluate headlines by error rates, baselines, logical versus physical qubits and verification.

Next: Project: A Grover SAT Solver