In October 2019, Google published a paper in Nature claiming that its Sycamore quantum processor had performed a specific computational task in 200 seconds that would take the world's most powerful classical supercomputer approximately 10,000 years. The announcement was accompanied by the phrase "quantum supremacy" — a term coined by physicist John Preskill in 2012 to describe the point at which a quantum computer can perform any computation that a classical computer cannot practically complete. The coverage was extensive, the language was dramatic, and the public understanding of what had actually been demonstrated was, in most cases, essentially wrong.
What Sycamore had done was genuinely impressive: it had performed a specific sampling task — drawing samples from a particular probability distribution defined by a random quantum circuit — faster than any known classical algorithm could. The task was designed to be hard for classical computers and natural for quantum ones. It was not designed to be useful. IBM, which had a competing interest in the matter, responded within days by arguing that its own classical supercomputer could complete the same task in 2.5 days using a different algorithm, not 10,000 years. The actual classical benchmark was contested. The underlying claim — that a quantum device had performed something no classical computer could practically do — rested on assumptions about the best available classical algorithms that were immediately challenged and have since been further revised.
None of this means quantum computing is not real, not advancing, or not eventually significant. It means that the public discourse around quantum computing has a systematic tendency toward exaggeration at the frontier and toward misrepresentation of what progress at the frontier actually implies for the applications that attract the most attention — encryption, drug discovery, artificial intelligence, financial modelling. Understanding what quantum computers can and cannot do, what the genuine technical obstacles are, and what realistic timelines look like requires separating a remarkable and genuinely important technology from the claims that have been attached to it.
What Quantum Computing Actually Is
Classical computers process information as bits — binary digits that are always in one of two states, 0 or 1. Every computation a classical computer performs, however complex, ultimately reduces to operations on strings of these binary values. The transistors that implement these operations switch between two electrical states, representing the two possible values of a bit. Modern processors contain billions of such transistors, switching billions of times per second, and the power of classical computing derives from this combination of scale and speed applied to well-understood binary logic.
Quantum computers use quantum bits, or qubits, which exploit properties of quantum mechanics that have no classical analogue. The most fundamental of these is superposition. While a classical bit is always 0 or 1, a qubit can exist in a superposition of both states simultaneously — not in the sense that it is randomly either 0 or 1, but in the strict quantum mechanical sense that it is in a linear combination of both states, with probability amplitudes for each outcome that can interfere with one another. When measured, the qubit collapses to a definite value of 0 or 1 with probabilities determined by its superposition state, but the computation happens in the superposition before measurement.
The second key property is entanglement. When two or more qubits are entangled, their states become correlated in ways that have no classical equivalent — measuring one qubit instantaneously determines information about the other, regardless of the physical distance between them. Einstein famously described this as "spooky action at a distance" and found it deeply troubling; subsequent experimental work by John Bell, Alain Aspect, and many others has established beyond reasonable doubt that quantum entanglement is a real physical phenomenon and not a hidden-variable artefact. For computation, entanglement allows quantum computers to represent and manipulate exponentially large state spaces using a number of qubits that grows only linearly with the problem size.
The third key property is interference. Quantum algorithms are designed to use interference — the constructive and destructive combination of probability amplitudes — to amplify the probability of correct answers and suppress the probability of incorrect ones. This is the mechanism by which quantum algorithms actually compute: not by trying all possible solutions simultaneously in some naive sense, but by carefully orchestrating interference patterns that guide the system toward correct outputs. Designing algorithms that achieve this is extraordinarily difficult, which is why the number of quantum algorithms with proven advantages over all known classical alternatives remains small.
Physical implementations of qubits use a range of quantum systems: superconducting circuits cooled to temperatures near absolute zero, trapped ions held in electromagnetic fields, photonic systems using individual photons, and semiconductor quantum dots, among others. Each approach has different strengths and weaknesses in terms of qubit coherence time, gate fidelity, connectivity, and scalability. No single implementation has yet demonstrated clear overall superiority, and different research groups and companies are pursuing different technological bets on which approach will prove most practical at scale.
Quantum Advantage: What It Means and When It Applies
The term "quantum advantage" — preferred by many researchers over "quantum supremacy" because of the latter's unfortunate connotations — refers to the situation where a quantum computer solves a problem faster than any classical computer can, given the best available algorithms for both. The definition sounds straightforward, but it contains several important qualifications that the popular discourse frequently elides.
First, quantum advantage is algorithm-specific and problem-specific. The quantum algorithms that have provably proven advantages over all classical alternatives are known for a relatively small set of problem types. Shor's algorithm, published in 1994, provides an exponential speedup for integer factorisation and discrete logarithm problems — the mathematical operations that underlie widely used public-key cryptographic systems including RSA and elliptic curve cryptography. Grover's algorithm provides a quadratic speedup for unstructured search problems. The variational quantum eigensolver and quantum phase estimation algorithms show theoretical promise for simulating quantum mechanical systems such as molecular structures. Beyond these, the catalogue of proven quantum advantages is considerably shorter than the popular discourse suggests.
Second, quantum advantage for a given algorithm does not automatically translate into practical utility. Shor's algorithm provides an exponential speedup for factorisation in theory, but running it on problem sizes large enough to threaten real-world encryption requires a fault-tolerant quantum computer with millions of error-corrected logical qubits — a capability that does not currently exist and that current technical trajectories suggest is at least a decade away, and possibly considerably longer. The theoretical advantage and the practical threat are separated by an enormous implementation gap.
Third, classical computing does not stand still. As quantum computing advances, so does classical computing, and as quantum researchers claim advantage on specific benchmarks, classical algorithm researchers develop better classical approaches that narrow or close the claimed gaps. The history of quantum supremacy claims since 2019 includes several instances where improved classical algorithms substantially reduced the apparent advantage within months of the original claim. This is not a reason to dismiss quantum computing, but it is a reason to treat specific benchmark comparisons as snapshots in a moving target rather than as definitive demonstrations.
The Error Problem: Why Current Quantum Computers Are Fragile
The most fundamental technical obstacle facing quantum computing at present is not the number of qubits — though scale matters — but qubit quality. Quantum states are extraordinarily fragile. Qubits interact with their environment through a process called decoherence, in which the quantum coherence of the superposition state is destroyed by thermal fluctuations, electromagnetic noise, mechanical vibrations, and the unavoidable imperfections of control systems. The time over which a qubit maintains useful quantum coherence — its coherence time — is typically measured in microseconds to milliseconds for current hardware, depending on the physical implementation.
Quantum gates — the operations that manipulate qubits during a computation — are also imperfect. Every gate operation introduces some small probability of error, and errors accumulate as computations lengthen. Current state-of-the-art two-qubit gate error rates are in the range of 0.1 to 1 percent per operation. This sounds small, but a computation requiring millions or billions of gate operations — the kind needed for practically useful quantum algorithms — would accumulate errors at a rate that destroys the computation long before it completes.
The solution to this problem is quantum error correction: using multiple physical qubits to encode each logical qubit in a redundant way that allows errors to be detected and corrected without destroying the computation. Theoretical quantum error correcting codes have been known since the 1990s, and recent years have seen important experimental progress in demonstrating that error correction can extend logical qubit coherence times beyond the coherence times of the physical qubits it is built from. Google's demonstration in 2023 that increasing the size of a surface code (a particularly practical error correction scheme) reduced the logical error rate — a phenomenon called "below-threshold" operation — was a genuine technical milestone, not a marketing claim.
The overhead of error correction is, however, enormous. Current estimates suggest that a single fault-tolerant logical qubit requires somewhere between hundreds and thousands of physical qubits to implement, depending on the target error rate and the physical qubit quality. A computation requiring a thousand logical qubits therefore requires millions of physical qubits. Current leading quantum computers have hundreds to a few thousand physical qubits of varying quality. The gap between current hardware and fault-tolerant computing at useful scale is real, large, and the central challenge of the field.
What Quantum Computers Can Actually Do Better
Setting aside the limitations and looking at where quantum computing has genuine promise, the most compelling near-term and medium-term applications cluster around a specific type of problem: the simulation of quantum mechanical systems.
Classical computers simulate quantum systems with exponential difficulty because the number of variables required to fully describe a quantum system grows exponentially with the system's size. Simulating the quantum mechanical behaviour of a molecule with 50 electrons requires keeping track of 2 to the power of 50 complex numbers — a quantity that exceeds the memory capacity of any classical computer for modest molecular sizes. Quantum computers, operating in a fundamentally quantum mechanical way, can represent quantum states naturally and efficiently. A quantum computer with 50 qubits can, in principle, represent the full quantum state of a 50-electron system with 50 qubits rather than with an astronomical number of classical bits.
The implications for chemistry, materials science, and drug discovery are significant. The electronic structure of molecules — which determines their chemical properties, reactivity, and biological activity — is a quantum mechanical problem that classical computers can only approximate. Better quantum simulation of molecular systems could enable more accurate prediction of drug-protein binding, more efficient discovery of catalysts for industrial chemistry, and better design of battery materials, superconductors, and other technologically important compounds. These are not abstract possibilities. Several pharmaceutical and chemical companies have active quantum computing research programmes targeting exactly these applications.
Optimisation problems — finding the best solution among a vast number of possible configurations — represent another category where quantum computers may offer advantages, though the evidence here is less clear than for quantum simulation. Problems like portfolio optimisation in finance, logistics network optimisation, and machine learning training involve searching large solution spaces in ways that might benefit from quantum approaches. The leading near-term approach is the Quantum Approximate Optimisation Algorithm, which runs on current noisy intermediate-scale quantum devices. Whether QAOA offers a genuine practical advantage over the best classical heuristics on practically relevant problem instances remains an open research question with no definitive answer yet.
What Quantum Computers Cannot Do
The gap between quantum computing's actual capabilities and the popular understanding of them is widest in two areas: encryption and general-purpose computing. Both deserve direct examination.
The claim that quantum computers will break all encryption is technically accurate in a narrow and practically irrelevant sense, and misleading in the sense that matters. Shor's algorithm does provide an exponential speedup for the specific mathematical problems that underlie RSA and elliptic curve cryptography — integer factorisation and discrete logarithm computation. A sufficiently large fault-tolerant quantum computer could, in principle, break these cryptographic systems. This is a real concern and has been taken seriously by the cryptographic community for decades.
The response has been commensurate. The US National Institute of Standards and Technology completed a multi-year competition in 2024 that standardised post-quantum cryptographic algorithms — mathematical problems believed to be hard for both classical and quantum computers — specifically to replace the quantum-vulnerable systems before fault-tolerant quantum computers exist. The winning algorithms, based on lattice problems and hash functions rather than factorisation and discrete logarithm problems, are already being integrated into protocols and standards. The cybersecurity community has a meaningful runway to complete this transition, and the work is actively underway. "Harvest now, decrypt later" attacks — where adversaries collect encrypted data now to decrypt once quantum computers exist — are a genuine concern for data that must remain secret for decades, but this is a specific and bounded threat, not the universal encryption apocalypse that popular accounts sometimes suggest.
More broadly, quantum computers are not general-purpose accelerators that are faster than classical computers at everything. They are specialised devices that offer advantages for specific problem types and are substantially inferior to classical computers for most of the computations that people actually need to perform. Running a word processor, browsing the internet, training a conventional neural network, playing a video game — none of these would benefit from quantum computation, and quantum hardware running these tasks would perform far worse than a modern laptop. The quantum computer is a specialist tool, not a universal replacement for the classical computer that sits on your desk.
The Hardware Landscape in 2025
The quantum computing hardware landscape in 2025 is characterised by genuine progress along multiple technological fronts, accompanied by intense commercial competition and a level of promotional activity that makes independent assessment difficult.
IBM has maintained the most systematic public roadmap of any major quantum computing company, targeting progressively larger and higher-quality systems under its Quantum System Two architecture. Its Heron processor, released in 2023, represented a significant improvement in two-qubit gate fidelity over previous generations. IBM's stated roadmap targets fault-tolerant quantum computing on a multi-year timeline, with intermediate milestones for demonstrating error-corrected logical qubits and small fault-tolerant computations.
Google Quantum AI has focused on demonstrating below-threshold error correction with its superconducting processors and has announced a roadmap targeting a useful fault-tolerant quantum computer within the decade. Its Willow chip, announced in late 2024, claimed significant improvements in error correction performance, though independent verification and the translation of chip-level benchmarks to application-level performance requires careful analysis.
Microsoft has pursued a distinct technical approach based on topological qubits — a type of qubit that is theoretically more resistant to environmental noise than the superconducting or trapped-ion qubits used by most competitors. Progress on topological qubits has been slower than Microsoft's earlier public timelines suggested, but recent publications have reported preliminary demonstrations of the underlying physical effects that topological qubits require.
Trapped-ion systems, pursued by companies including IonQ and Quantinuum (formerly Cambridge Quantum and Honeywell Quantum Solutions), offer different trade-offs: typically higher gate fidelities than superconducting systems but slower gate speeds and more complex scalability paths. Quantinuum's H-series processors have demonstrated some of the highest two-qubit gate fidelities of any commercially available system, and the company has published detailed technical results on small-scale error correction demonstrations.
Photonic quantum computing, pursued by companies like PsiQuantum and Xanadu, offers the theoretical advantage of room-temperature operation and natural integration with optical telecommunications infrastructure, but faces significant engineering challenges in achieving the low photon loss rates and deterministic photon-photon interactions that large-scale computation requires.
Realistic Timelines
Forecasting timelines in a rapidly evolving technical field is inherently uncertain, and the track record of quantum computing timeline predictions has not been impressive. John Preskill, who coined the term "quantum supremacy" and is among the most respected voices in the field, has consistently cautioned against specific near-term predictions and has emphasised that the engineering challenges of fault-tolerant quantum computing are substantially harder than the theoretical physics of quantum algorithms. His framing — that the question is not if but when useful fault-tolerant quantum computing will exist, with "when" potentially being decades rather than years — is probably the most calibrated available from a credible source.
The research community's consensus, as best as it can be characterised from published literature and conference discussions, suggests the following rough framework. Noisy intermediate-scale quantum (NISQ) devices — current quantum computers with tens to hundreds of qubits and without full error correction — may demonstrate genuine practical advantages over classical computers in specific narrow applications within the next three to five years, most likely in quantum chemistry simulation for small molecules. This would be a meaningful milestone, though it would not represent broad commercial utility.
Small-scale fault-tolerant quantum computing — systems with tens to hundreds of logical error-corrected qubits capable of running quantum algorithms of modest depth — is most plausibly a decade away, with considerable uncertainty in both directions. Large-scale fault-tolerant quantum computing capable of running Shor's algorithm on cryptographically relevant key sizes is, on most credible assessments, fifteen to thirty years away, possibly longer. Anyone claiming a specific year for these milestones with high confidence is expressing a level of certainty that the technical evidence does not support.
What This Means for Investment and Policy
The gap between quantum computing's current state and the transformative applications it is most frequently associated with in policy and investment discussions has practical consequences for how resources should be allocated.
Government investment in quantum computing research is justified on multiple grounds: the genuine long-term promise of the technology, the national security implications of quantum-vulnerable cryptography, and the competitive dynamics of a field where several major economies are making substantial state investments. The US National Quantum Initiative, the EU Quantum Flagship programme, and equivalent initiatives in China, the UK, and elsewhere represent a recognition that quantum computing is a strategically important technology worth sustained public investment. These investments are broadly defensible on the evidence.
Private investment is a more complex picture. The quantum computing startup ecosystem has attracted billions of dollars in venture capital on timelines and use-case projections that, in many cases, outrun the technical evidence. Companies claiming near-term commercial advantage in drug discovery, financial optimisation, or logistics on current NISQ hardware are making claims that require careful scrutiny. The genuine near-term commercial opportunity in quantum computing is probably more modest than the funding landscape suggests — concentrated in quantum sensing, quantum networking, and narrow quantum simulation applications rather than in the broad across-the-board performance advantages that investor pitches tend to project.
For organisations making technology procurement or cybersecurity decisions, the most actionable implication of the current quantum computing landscape is not to rush quantum deployment but to begin planning for post-quantum cryptographic migration. The timeline for cryptographically relevant quantum computers is long but not infinite, and the migration of large-scale IT infrastructure to post-quantum standards takes years. Starting that planning process now — inventorying which systems use quantum-vulnerable cryptography, prioritising the most sensitive data stores for early migration, and tracking the NIST post-quantum standards as they are finalised and implemented — is a reasonable and well-supported recommendation regardless of one's view on the precise quantum computing timeline.
A Technology Worth Understanding Honestly
Quantum computing is a genuine scientific and engineering achievement that has progressed substantially over the past two decades, and the pace of progress continues to accelerate. The underlying physics are not in doubt. The demonstrations of specific quantum mechanical effects in engineered systems are real and have been independently verified. The theoretical advantages of quantum algorithms for specific problem classes are mathematically proven. And the engineering progress toward larger, higher-quality quantum hardware is measurable and documented.
What quantum computing is not is the imminent universal revolution that a naive reading of press releases and investment materials might suggest. The gap between current hardware and practically useful fault-tolerant quantum computing is large, the engineering challenges are formidable, and the set of problems where quantum computers will offer real advantages over the best classical alternatives is narrower than the popular account implies. These are not reasons for pessimism. They are reasons for intellectual honesty about a technology whose genuine promise is remarkable enough that it does not require inflation to be worth taking seriously.
The most useful frame for a non-specialist trying to navigate quantum computing claims is probably this: when you encounter a headline announcing a quantum computing breakthrough, ask what specific problem was solved, what the best classical alternative performance was, whether the comparison was on problem instances of practical relevance, and whether independent researchers have confirmed the result. Most of the time, the honest answers to those questions will reveal something more modest than the headline, and occasionally they will reveal something genuinely important. Distinguishing between the two is the most valuable skill a scientifically engaged reader can bring to this field — and it is a skill that serves equally well everywhere that hype and reality are running on different tracks.