Safekipedia

Quantum computing

Adventurer · ages 7–10Adapted from Wikipedia
An IBM Quantum System One, a quantum computer, on display in Ehningen, Germany.

A quantum computer is a computer that stores and works with information using quantum states. It uses effects such as superposition, interference, and entanglement. Someday, quantum computers could finish some calculations much faster than ordinary computers. For example, a large quantum computer could break widely used encryption schemes and help physicists with physical simulations. But today's quantum computers are still mostly experimental and are only useful for certain special tasks.

The basic unit of information in quantum computing is the qubit, or quantum bit. It does a job similar to the bit in ordinary, or "classical," computers. A classical bit can be in one of two states, a binary value. A qubit can exist in a linear combination of states called a quantum superposition. When a qubit is measured, the result is one of the two states, chosen by a probabilistic rule. If a quantum computer handles the qubit in a special way, wave interference effects can make the desired result more likely. Quantum algorithm design means creating step-by-step procedures that help a quantum computer do this.

IBM quantum computer demo at ITU WTSA 2024 in Delhi

Quantum computers are not yet practical for everyday use. If a physical qubit is not well isolated from its surroundings, it suffers from quantum decoherence, which adds noise, or errors, to calculations. Governments have funded research to build qubits that keep their quantum states longer and make fewer errors. Some designs use superconductors, which carry an electrical current with no electrical resistance. Others use ion traps, which hold a single atomic particle in place using electromagnetic fields.

Researchers have claimed that quantum devices can beat classical computers on certain tasks, a goal called quantum advantage or quantum supremacy. These tasks are not always useful in the real world. So, as of 2026, these demonstrations are best seen as scientific milestones, not proof that quantum computers will be ready for use soon. Global government investment in quantum computing reached $10 billion by April 2025.

History

For a year-by-year guide, see Timeline of quantum computing and communication. Quantum mechanics and computer science were separate fields of study until quantum computing came along. Quantum theory was developed in the 1920s to explain strange things seen in physics. Computers came decades later. As physicists began using quantum ideas for computing problems and replacing bits with qubits, the two fields started to join together. In 1980, Paul Benioff introduced the quantum Turing machine, which used quantum theory to describe a simple computer. As digital computers grew faster, the effort needed for simulating quantum dynamics still grew in an exponential way. So Yuri Manin and Richard Feynman each suggested that computers based on quantum effects might work better.

Peter Shor (pictured here in 2017) showed in 1994 that a scalable quantum computer would be able to break RSA encryption.

In 1984, Charles Bennett and Gilles Brassard used quantum theory to improve secret communication. They showed that quantum key distribution could help protect information security.

Quantum algorithms came next, designed for oracle problems. These included Deutsch's algorithm in 1985, the Bernstein–Vazirani algorithm in 1993, and Simon's algorithm in 1994. These methods did not solve practical problems. However, they showed on paper that more information could be found by asking a black box questions using a quantum state in superposition, which is sometimes called quantum parallelism.

In 1994, Peter Shor built on these ideas with polynomial-time quantum algorithms for integer factorization and the discrete logarithm problem. A large enough quantum computer could break some public-key cryptography systems that are widely used. Fast factorization would weaken RSA, and a fast discrete-logarithm method would weaken Diffie–Hellman key exchange. This attention to Shor's algorithm brought wide interest to quantum computing. In 1996, Grover's algorithm showed a quantum speedup for the unstructured search problem. That same year, Seth Lloyd showed that quantum computers could simulate quantum systems without the extra work that classical methods need. This supported Feynman's idea from 1982. Experimentalists built small quantum computers using trapped ions and superconductors. In 1998, a two-qubit machine showed that the idea could work in practice.

In 2019, Google AI and NASA announced that they had reached quantum supremacy with a 54-qubit machine. They said it finished a task that would take a classical supercomputer a very long time. IBM disagreed and said the task could be done in about 2.5 days on its Summit supercomputer. This led to debate about what counts as this milestone. Later work focused on quantum error correction, which helps control decoherence. In 2024, researchers showed progress toward large, reliable quantum memory. This is a step toward bigger systems, but building them at scale is still a challenge for engineers.

See also: Unitarity (physics)

Further information: Quantum programming

Quantum information processing

A quantum circuit diagram implementing a Toffoli gate from more primitive gates

Computer engineers usually describe how a modern computer works using classical physics. Some parts of these computers may depend on quantum behavior, but they are not fully cut off from their surroundings. Because of this, any quantum information quickly decoheres, or falls apart. Classical computers store information in bits that are either 0 or 1. The qubit is the basic unit of quantum information. When measured, a qubit gives either 0 or 1. Before that, it can be in a superposition of both. Each added qubit doubles the size of the system, so a classical computer finds it hard to copy a quantum one. For example, a 100-qubit system needs 2100 values to be stored. Quantum programs depend on carefully controlling quantum systems. These systems are described with complex numbers, vectors, and matrices. Quantum computers use quantum logic gates, such as the CNOT gate, to change qubits. This works much like classical logic gates change bits. Physicist Charlie Bennett pointed out that classical computers are made of quantum atoms. So all computers are quantum, and the real question is where classical slowdowns come from.

Any quantum computation can be built from a small set of gates. This is called a universal gate set. Researchers also describe computing as a network of gates and measurements. These are often drawn as quantum circuits. Quantum computers are still being developed. As of 2026, fully error-free machines are still out of reach. Some researchers think noisy intermediate-scale quantum machines may soon have special uses. However, errors in gates limit how reliable they are. Scientists at Harvard created quantum circuits that fix errors more efficiently than other methods. This may help remove a major obstacle. The Harvard team was supported by MIT, QuEra Computing, Caltech, and Princeton. The work was funded by DARPA's Optimization with Noisy Intermediate-Scale Quantum devices (ONISQ) program.

Communication

Further information: Quantum information science

Quantum cryptography makes safe data sending possible. For example, quantum key distribution uses linked quantum states to create secure cryptographic keys.: 1017 If a sender and receiver pass quantum states between them, they can tell if an adversary has listened in, because spying would disturb the fragile quantum system and cause a change that can be noticed. Using suitable cryptographic protocols, they can then share information privately. Modern fiber-optic cables can carry quantum information over fairly short distances. Researchers are working on more reliable parts, such as quantum repeaters, to make long-distance quantum networks possible. In theory, this could allow new uses, such as distributed quantum computing and better quantum sensing.

Quantum teleportation is a method where Alice can send the state of a qubit to Bob. She needs one shared entangled pair and two classical bits. Alice's qubit is not physically sent. Instead, it is rebuilt at Bob's end using measurement results sent over ordinary channels. This shows that quantum communication needs both entanglement and regular communication. Teleportation cannot send information faster than light. Superdense coding works the other way. Using one shared entangled pair and sending only one qubit, Alice can send two classical bits to Bob. This seems to break Holevo's theorem, which says a single qubit can carry at most one bit of classical information. The shared entanglement is what makes this possible.

Algorithms

See also: List of quantum algorithms

Most work on quantum algorithms uses the quantum circuit model. A few exceptions exist, such as the quantum adiabatic algorithm. Scientists group quantum algorithms by the kind of speedup they give over regular, or classical, algorithms. Some give more than a polynomial speedup, including Shor's algorithm for factoring. Related methods work for discrete logarithms, Pell's equation, and the hidden subgroup problem for abelian finite groups. They all use the quantum Fourier transform. No one has proven that a classical algorithm could never be just as fast, but the evidence suggests this is unlikely. Some problems, such as Simon's problem and the Bernstein–Vazirani problem, show clear speedups. However, these speedups only apply in the quantum query model, a restricted model that does not always match real problems.

Other problems appear to have very large speedups. These include simulating quantum physical processes from chemistry and solid-state physics, approximating certain Jones polynomials, and the quantum algorithm for linear systems of equations. They are BQP-complete. Some algorithms, such as Grover's algorithm and amplitude amplification, give smaller polynomial speedups but can be used for many problems. Their speedups have not yet been shown in real-world use.

Simulation of quantum systems

Main article: Quantum simulation

Chemistry and nanotechnology rely on understanding quantum systems. These systems are very hard to simulate on regular computers, so quantum simulation may be an important application. Quantum computational chemistry is promising, though useful versions are still limited by hardware. Quantum simulation could also show how atoms and particles behave under unusual conditions, such as reactions inside a collider. In June 2023, IBM computer scientists reported that a quantum computer gave better results for a physics problem than a conventional supercomputer.

About 2% of the world's yearly energy is used for nitrogen fixation to make ammonia for the Haber process, which is used to make fertiliser for farms. Quantum simulations might help scientists understand this process and make it more energy efficient.

Cryptography

Main articles: Post-quantum cryptography and Quantum cryptography

Digital cryptography keeps communications private by stopping unauthorised people from reading them. Ordinary encryption hides a message with a key, using a process that is hard to reverse. Quantum computing may someday be powerful enough to make these reversals possible. In theory, it could attack many cryptographic systems used today. Integer factorization protects public key cryptographic systems. It is believed to be too hard for a classical computer when the number is the product of a few large prime numbers. A quantum computer could solve this much faster using Shor's algorithm, which could break many widely used systems. These include RSA, Diffie–Hellman, and elliptic curve Diffie–Hellman, which help keep web pages, emails, and other data safe.

Researchers are working on post-quantum cryptography, which aims to find systems that quantum algorithms cannot easily break. Some public-key methods, such as the McEliece cryptosystem, rely on problems that Shor's algorithm cannot solve. Lattice-based cryptosystems are not known to be vulnerable to quantum computers. Grover's algorithm can also weaken symmetric (secret-key) algorithms, but doubling the key length keeps them about as secure. Quantum cryptography uses ideas from quantum mechanics, such as entanglement, to protect messages. In principle, a message protected this way cannot be decoded even by a quantum computer. However, this takes a lot of special equipment.

A wafer of adiabatic quantum computers

Search problems

Main article: Grover's algorithm

The best-known example of a problem with a polynomial quantum speedup is searching an unstructured list to find one marked item. Grover's algorithm can do this with fewer checks than classical methods. Many other speedups are based on it, including Brassard, Høyer, and Tapp's algorithm for finding collisions. Grover's algorithm can be used on a Boolean satisfiability problem. It could also be used in a password cracker that guesses passwords. Breaking symmetric ciphers this way is of interest to government agencies.

Quantum annealing

Quantum annealing uses the adiabatic theorem to do calculations. The system starts in a simple state and slowly changes into a more complex one. Its lowest state represents the answer. If the change is slow enough, the system stays in that lowest state. Quantum annealing can solve Ising models and related problems, which can represent many combinatorial optimization problems. It may also help with computational biology problems.

Machine learning

Main article: Quantum machine learning

Quantum computers may be able to speed up some machine learning tasks. However, many proposed advantages depend on special assumptions. As of 2024, they had not led to a clear practical benefit. The HHL Algorithm is believed to offer a speedup over classical methods. Researchers have also tried quantum annealing hardware for training Boltzmann machines and deep neural networks.

Quantum methods are also being explored for drug discovery. In 2023, researchers reported a hybrid model that generated new small molecules with properties similar to medicines. Large molecule spaces remain a major challenge, which future quantum computers might help with. Quantum-enhanced models such as quantum generative adversarial networks (GANs) may be developed into chemistry tools.

AI-assisted algorithm discovery

Artificial intelligence is being explored as a tool for finding and improving quantum algorithms. AlphaEvolve, a Google DeepMind system built on large language models and evolutionary algorithms, has been described as a coding agent for scientific discovery. In quantum research, AlphaEvolve-optimised quantum circuits have been used in work on quantum computation of molecular geometry through many-body nuclear spin echoes.

Engineering

See also: Quantum network

Further information: List of proposed quantum registers

Quantum System One, a quantum computer by IBM from 2019 with 20 superconducting qubits

As of 2023, regular computers were still better than quantum computers for every real-world task. Building a large quantum computer is very hard. Physicist David DiVincenzo listed what a useful one needs. It needs to be able to add more qubits. Its quantum decoherence needs to be slower than its operations. Its qubits also need to be easy to read. Parts are hard to get, too. Superconducting quantum computers need helium-3, which is hard to find. Errors are another challenge, so scientists study quantum error correction and topological quantum computers. The National Institute of Standards and Technology (NIST) has also started work on codes that stay safe from both quantum and regular computers.

In 2019, Google Quantum AI said it had reached quantum supremacy. This means a quantum device solved a problem that regular computers could not solve well. Other researchers questioned this claim. A 2023 article said that quantum computers were not yet useful, but they might be useful one day. Among the most developed types of hardware are superconductors and trapped ions.

Theory

Computability

Further information: Computability theory

The suspected relationship of BQP to several classical complexity classes

Any computational problem that a classical computer can solve can also be solved by a quantum computer. This is because everything in the physical world, including classical computers, can be described using quantum mechanics. That same idea explains how quantum computers work. In the other direction, any problem a quantum computer can solve can also be solved by a classical computer, if it is given enough time. Quantum computers are not more powerful than classical ones. They cannot solve undecidable problems like the halting problem. Their existence also does not disprove the Church–Turing thesis.

Complexity

Main article: Quantum complexity theory

Quantum computers cannot solve any problem that classical computers cannot already solve. However, they are thought to solve some problems faster. For example, quantum computers can efficiently factor integers, which classical computers are not believed to do well. The problems a quantum computer can efficiently solve with bounded error form the class BQP. BQP is the quantum version of BPP. Scientists do not yet know exactly how BQP relates to PSPACE and other classes. They do know that any problem classical computers can solve efficiently can also be solved efficiently by quantum computers. Integer factorization and the discrete logarithm problem are believed to be solvable by quantum computers but not by classical ones. If a quantum computer could solve an NP-complete problem efficiently, then, because of NP-hardness, it could also solve all problems in NP efficiently. This is not believed to be true.

List of quantum computers

See also: List of quantum processors

Some quantum computers have already been built. The Hanyuan-1 is a 100-qubit neutral atom quantum computer made by the Chinese Academy of Sciences in China. The IBM Quantum System One is an IBM superconducting computer that was introduced in 2019. The IBM Quantum System Two is a modular superconducting computer that uses IBM Heron processors. Jiuzhang is a photonic prototype built for Gaussian boson sampling. The QpiAI-Indus is a 25-qubit superconducting quantum computer from QpiAI in India.

Scientists are also studying many other ways to build quantum computers. Some of these ideas have only been proposed so far. One is the Cat qubit quantum computer, which is based on cat-state qubits. Another is the Kane quantum computer, which uses a silicon design with nuclear spin. Other ideas use photonic methods. One example is linear optical quantum computing, which works with photons. Some approaches use neutral atoms that are held and controlled with light. Others use nuclear magnetic resonance or superconducting electronic circuits. The Spin qubit quantum computer uses a semiconductor design. The Topological quantum computer is proposed to use states such as anyons. The Trapped-ion quantum computer uses charged atoms that are held in place as qubits.

Quantum computing conferences

  • Quantum Information Processing (QIP)
  • Theory of Quantum Computation, Communication and Cryptography (TQC)
  • Quantum Communication, Measurement and Computing (QCMC)
  • IEEE International Conference on Quantum Computing and Engineering (QCE), also known as IEEE Quantum Week
  • International Conference on quantum cryptography (QCrypt)
  • Q2B – by QC Ware.

Quiz time!

4 questions about Quantum computing

Question 1 of 4

What does a quantum computer use to store and work with information?

Keep exploring

This article is a child-friendly adaptation of the Wikipedia article on Quantum computing, available under CC BY-SA 4.0.

Images from Wikimedia Commons. Tap any image to view credits and license.