What quantum technology is, why it exists, where it already works, what it may do next, and how to learn it from basic to advanced. Every concept has a long explanation and a live simulation you can break and rebuild.
Drag the scene to rotate it.
People use the words as if they meant the same thing. Separating them makes everything else on this page easier to follow.
The rules that govern matter and light at the scale of atoms and below. Energy comes in discrete packets called quanta. Particles also behave like waves. A system can be in a superposition of several states until it is measured, and the result of a measurement is random, with probabilities set by the state.
These rules have been tested for a century. They already underpin transistors, lasers, LEDs, solar cells, MRI scanners and atomic clocks. That is often called the first quantum revolution.
The second quantum revolution. Instead of using quantum effects as a side effect inside bulk materials, engineers now build and control individual quantum systems on purpose: single atoms, ions, photons, electron spins and superconducting circuits.
Four families result: computing, communication, sensing and simulation. They share the same ingredients (superposition, interference, entanglement) and the same enemy (noise from the environment).
One branch of quantum technology. A quantum computer stores information in qubits and runs algorithms that steer probability amplitudes so that wrong answers cancel and the right answer is likely to be measured.
It is not a faster laptop. It is a different kind of machine that gives large advantages on some problems, small ones on others, and none on most everyday tasks.
| Property | Classical bit | Qubit |
|---|---|---|
| State | 0 or 1 | α|0⟩ + β|1⟩ with complex α, β and |α|² + |β|² = 1 |
| n of them | One of 2ⁿ values at a time | 2ⁿ complex amplitudes at once, but reading gives only n bits |
| Reading | Does not disturb the value | Measurement is random and collapses the state |
| Copying | Free | Impossible for an unknown state (no-cloning theorem) |
| Operations | Logic gates, many irreversible (AND, OR) | Unitary gates, always reversible |
| Correlation | Shared randomness | Entanglement, which can violate Bell inequalities |
| Errors | Around 10⁻¹⁷ per operation in a modern chip | Around 10⁻³ to 10⁻² per operation on today’s devices |
| Hardware | Silicon transistors at room temperature | Superconducting circuits, ions, atoms, photons or spins, often near absolute zero or in vacuum |
Quantum technology is not built for speed in general. It is built for four specific jobs.
1. Simulate nature. In 1981 Richard Feynman pointed out that a classical computer needs exponentially more resources to track a quantum system, so the natural simulator is another quantum system. Molecules, catalysts, magnets and superconductors are the target.
2. Solve certain mathematical problems with fewer steps. Shor’s algorithm factors large integers in polynomial time. Grover’s algorithm searches unstructured data with a quadratic speedup. Other algorithms help with linear algebra, sampling and estimating averages.
3. Measure more precisely. A quantum probe such as an atom or a spin is extremely sensitive to magnetic fields, gravity, rotation and time. Quantum sensors already tell time for satellite navigation and image brain activity.
4. Communicate with physics-backed security. Measuring an unknown quantum state disturbs it, so eavesdropping on a quantum key exchange leaves a detectable trace. Quantum networks will also link quantum computers together.
The best known classical algorithm for factoring, the general number field sieve, takes time that grows faster than any polynomial in the number of digits. Shor’s algorithm grows roughly as the cube of the number of bits. The chart shows the shape of that gap on a log scale. It ignores constant factors and the large overhead of error correction, so read it as a statement about how the cost grows, not as a prediction of run time.
For unstructured search the gain is smaller. Grover gives a square-root speedup, which turns 10¹² steps into 10⁶ queries but does not make hard problems easy.
Computing gets the headlines, but communication and sensing are further along commercially.
Qubits plus gates plus measurement. Three models exist: gate-based (circuits, the one on this page), annealing (optimisation by slowly changing a Hamiltonian) and analog (a device that directly mimics one physical system).
Quantum key distribution (BB84, E91) shares encryption keys whose secrecy rests on physics. Quantum networks send entangled photons between nodes, and repeaters will extend range beyond the roughly one hundred kilometres loss allows in fibre.
A superposition accumulates phase in proportion to a field or to elapsed time. Reading that phase gives extreme sensitivity. Atomic clocks, diamond NV magnetometers, atom interferometers and squeezed-light detectors all work this way.
Analog simulators use cold atoms in optical lattices, Rydberg atom arrays or trapped ions as a stand-in for a material you cannot compute. Digital simulation uses a gate-based computer to evolve the same model step by step.
It is classical mathematics, run on ordinary computers, designed to resist attacks from quantum computers. NIST published its first three post-quantum standards in 2024: ML-KEM (FIPS 203) for key exchange, and ML-DSA (FIPS 204) and SLH-DSA (FIPS 205) for signatures. Most organisations will migrate to these long before they ever buy a QKD link, because they run on existing networks and hardware. The two approaches are complementary.
Nobody runs a whole application on a quantum chip. A classical program prepares the problem, calls the quantum processor for the one step it does well, and processes the answer.
Several vendors offer cloud access to real quantum processors, including IBM Quantum, Amazon Braket, Azure Quantum and the cloud services of ion-trap, neutral-atom and photonic companies. Free tiers or credits exist on some of them, so check current terms. All the major SDKs also ship simulators that run on a laptop.
A sensible path is to write circuits for 2 to 5 qubits, run them on a simulator first, then run the same circuit on real hardware and compare the two histograms. The difference between them is a direct picture of noise.
Quantum sensors and QKD systems are bought as instruments. They plug into existing navigation, imaging or network equipment and need no quantum programming.
No platform has won. Each trades speed, fidelity, connectivity and manufacturability differently, and the field keeps changing, so treat the company names as examples rather than a ranking.
A small circuit with a Josephson junction behaves as an artificial atom. Chips are cooled to about 10 to 20 millikelvin and driven with microwave pulses.
Individual charged atoms are held in electric fields and laser-cooled. Their internal energy levels are the qubit, and lasers or microwaves drive the gates.
Atoms such as rubidium or strontium are held in arrays by focused laser beams called optical tweezers. Exciting them to Rydberg states makes neighbours interact and entangle.
Photons carry the qubit in their path, polarisation, time bin or in squeezed modes of light. Circuits are waveguides on a chip with beam splitters and phase shifters.
The spin of a single electron, or of a donor nucleus, is trapped in a transistor-like structure and controlled with microwaves and voltages.
A nitrogen atom next to a missing carbon atom (an NV centre) has a spin that can be initialised and read with light, even at room temperature.
Superconducting flux qubits implement an Ising-type Hamiltonian that is slowly changed so the system settles into a low-energy solution of an optimisation problem.
An idea in which information lives in non-local properties of exotic quasiparticles, so local noise cannot easily damage it.
The tag on each card says how close the application is to everyday use. Several well-known promises are still research.
Molecular energies are quantum problems. Accurate simulation of reaction mechanisms, binding and enzymes is limited by classical approximations. A widely studied target is the iron-molybdenum cofactor of nitrogenase, which fixes nitrogen in nature.
Electrolytes, catalysts, magnets and high-temperature superconductors are hard because electrons interact strongly. Simulating them could guide the search for better batteries and cheaper catalysts.
A large fault-tolerant machine running Shor’s algorithm would break RSA and elliptic-curve cryptography. Data stolen today can be decrypted later, so migration is already under way.
Routing, scheduling, portfolio selection and supply chains are combinatorial. Annealers and QAOA are tried on them by airlines, ports, banks and manufacturers.
Pricing derivatives and measuring risk rely on Monte Carlo sampling. Amplitude estimation offers a quadratic reduction in the number of samples needed for a given accuracy.
Ideas include quantum kernels, variational classifiers and faster sampling for generative models. Quantum data from sensors and simulators is a more natural fit than classical images or text.
Fertiliser production by the Haber–Bosch process consumes on the order of one to two percent of the world’s energy. Better catalysts, carbon-capture materials and fusion plasma models are all quantum-simulation targets.
Atomic clocks keep satellite navigation and telecom networks synchronised. Optical clocks are far more precise, and cold-atom accelerometers and gyroscopes can navigate when GPS is unavailable.
Optically pumped magnetometers sit in a wearable helmet and record the tiny magnetic fields of brain activity without a bulky cryogenic scanner. Diamond sensors probe single cells and molecules.
Quantum links will secure key exchange and eventually connect quantum computers into larger distributed machines. Repeaters with quantum memories are the key missing component.
Read the text on the left and play with the simulation on the right. Each lesson says what the idea is, how it works, where it is used and what to try. They run roughly from basic to advanced, and each is tagged with its level.
The history shows how slowly foundations turn into products. The roadmap shows what hardware companies and standards bodies are planning.
Accurate simulation of how a drug binds to a protein, including quantum effects that classical force fields miss, could shorten the search for candidates.
Room-temperature superconductors, better magnets, lighter and safer batteries and efficient solar absorbers are all searches through quantum-mechanical possibility space.
Catalysts for fertiliser, hydrogen and carbon capture could cut a large share of industrial energy use if simulation can guide their design.
Plasma dynamics and materials for reactor walls are very hard to simulate. Quantum algorithms for differential equations are an active research area.
Entanglement distribution between cities would allow secure communication, networked sensors with higher precision, and cloud access to quantum computers with private data.
Linking many smaller processors with photonic interconnects may scale further than a single chip can.
Compact cold-atom inertial sensors could keep ships, aircraft and vehicles on course for hours without satellite signals.
Gravimeters for mapping aquifers and tunnels, and magnetometers that read brain and heart signals without shielded rooms.
A register of n qubits holds 2ⁿ amplitudes, but a measurement returns only one n-bit result. Algorithms work by interference that makes the right answer likely, not by reading all branches.
Quantum computers are slow at ordinary tasks and need delicate, expensive infrastructure. They will sit in data centres as accelerators for a few kinds of work.
Breaking RSA-2048 needs a large fault-tolerant machine that does not exist yet. Symmetric encryption such as AES-256 is only weakened by a square root. Migration to post-quantum standards is under way.
Most problems have no known quantum speedup. Many known ones are only quadratic, and that is often not enough to cover the overhead of error correction.
Measurement results on each side look random. They only show a correlation once the two sides compare notes by ordinary communication, which is no faster than light.
Quality matters as much as count. Gate fidelity, connectivity, coherence, readout and the ability to run error correction decide what a machine can do.
Tick topics off as you finish them. Progress is saved in this browser only.
CNOT rows and columns are ordered |00⟩, |01⟩, |10⟩, |11⟩ with the control qubit written first. H, S and CNOT generate the Clifford group. Adding T makes the set universal.
Pick a level or take them all. Every answer comes with an explanation. Scoring 80% or more is a good sign you can move on.
These are ideas to pick from, not step-by-step tutorials. Finish one project per level and you will have a portfolio that shows real depth.
Pick one SDK and stay with it for a while. The concepts transfer between all of them.
Twenty-six step-by-step modules, an immersive 3D universe, eleven hands-on challenges, a roadmap, glossary, quiz and projects.
Twenty-six modules, fifteen live simulations, an interactive Quantum Lab with eleven challenges, an immersive 3D universe, a roadmap, glossary, quiz and twelve projects – computing, communication, sensing, hardware and the future, explained step by step.
Start learningEnter the Quantum LabSee projectsEverything you learned above is now yours to try. Set the inputs, run the experiment, read the results and complete the challenges. Progress: 0 / 11 challenges completed.
Amplitude complex number whose squared size is a probability · Basis reference set of states · BQP problems quantum computers solve efficiently · Coherence preserved phase relations · Decoherence loss of coherence to the environment · Fidelity overlap between states or accuracy of an operation · Hamiltonian energy operator governing evolution · Hilbert space state space of a system · Logical qubit error-protected qubit built from many physical ones · NISQ noisy intermediate-scale quantum era · Observable measurable quantity · Syndrome error-revealing parity measurement · Tomography reconstructing a state from measurements · Transpilation compiling a circuit to hardware · Unitary reversible, probability-preserving operation · Qudit d-level generalisation of a qubit · Clifford gates efficiently simulable gate set (H, S, CNOT) · Magic state resource enabling non-Clifford T gates.
Progress from beginner to research-grade. Free tools: Qiskit, Cirq, PennyLane, IBM Quantum, Amazon Braket.
What IBM, Microsoft, Google and Qiskit offer, a live study planner, a 3D Qiskit stack and five extra simulations.
Four of the main places to learn quantum technology for free or to run real experiments. This is a summary of what each currently offers, with links so you can open the originals.
A library of self-paced courses by IBM and outside experts. It starts with a quick-start path, then a four-course foundations series taught by John Watrous, then applied courses and two business courses.
A beginner-to-advanced learning path for developers. You learn the qubit, write Q# programs with the Quantum Development Kit, then run on real hardware through the Azure Quantum cloud.
Prerequisites: basic Azure knowledge, basic linear algebra, familiarity with Visual Studio Code. Open the path
Google builds superconducting quantum chips and open software. Its site covers hardware, a roadmap, the Cirq library, research papers and learning resources.
Qiskit builds, optimises and runs quantum workloads. Install it with pip install qiskit. The site announces release 2.6 and lists these headline figures (IBM's own claims, or citing a Unitary Foundation survey and the Benchpress toolkit):
Tick the courses you want. The total hours update live. Filter by group.
No courses selected.
At 5 hours a week this is 0 weeks.
Click a layer, or hover over it, to see the tools. The stack floats and tilts on its own.
Quantum Approximate Optimization Algorithm (QAOA) for combinatorial problems.
Sample-based quantum diagonalization of a chemistry Hamiltonian, using quantum and HPC together.
Quantum kernel training to boost classical machine learning.
Modelling a flowing non-viscous fluid with variational quantum methods (QUICK-PDE).
Qiskit has a Slack community, events such as Fall Fest, an ecosystem of community extensions and an advocate programme with well over a thousand members in more than 70 countries.
A superconducting processor built to show that adding qubits to an error-correcting code can reduce errors, a key step toward large fault-tolerant machines.
An algorithm on Willow that Google describes as the first verifiable quantum advantage, meaning the result can be checked by repeating it on another quantum computer.
A method for preparing the special states that give error-corrected machines their non-Clifford gates, demonstrated on Willow.
Google now describes work on both superconducting and neutral-atom quantum computers.
A call for proposals for researchers who want to run experiments on Willow.
Google's open-source Python library for writing, simulating and running quantum circuits.
All of them run in your browser. Drag the 3D views to rotate them.
Pick a qubit, add gates, and watch the 3D probability bars and the shot counts change.
Two qubits that are not neighbours need SWAP gates. See the route and the fidelity cost.
The quantum walker spreads in proportion to the number of steps, the classical one only to its square root.
Rough surface-code estimate in the spirit of the Microsoft resource estimator. Model: threshold 1%, logical error 0.1·(p/0.01)^((d+1)/2), 2(d+1)² physical qubits per logical qubit, 1 µs per round.
Choose a goal and get an ordered list from the IBM, Microsoft, Google and Qiskit resources above.