What You Need to Know About Quantum Computing

<p>I spent a decent chunk of the last year trying to separate what quantum computing actually is from what the press releases claim it is, and the gap turned out to be enormous. Every few months a headline announces that quantum computers will shatter all encryption, cure disease, or break the laws of economics by next Tuesday. Almost none of it survives contact with how the machines really work. So here is the version I wish someone had handed me at the start: what a quantum computer genuinely does, why the physics is so strange, where the real breakthroughs have been, and which of the promises you can safely ignore for now.</p> <h2 id="why-classical-computers-have-a-ceiling">Why Classical Computers Have a Ceiling</h2><div class="ad-unit ad-in-article" aria-label="Advertisement"> <span class="ad-label">Advertisement</span> <ins class="adsbygoogle" style="display:block;text-align:center" data-ad-client="ca-pub-3726833845844946" data-ad-slot="3291553914" data-ad-format="auto" data-full-width-responsive="true"></ins> <script>(adsbygoogle = window.adsbygoogle || []).push({});</script> </div> <p>Everything you have ever run — this page, your operating system, the servers behind your favourite app — is built on bits. A bit is either 0 or 1, and everything else is combinations of billions of them flipping very quickly. Classical computers are astonishingly good at this, and for the overwhelming majority of tasks they are all you will ever need.</p> <p>But some problems scale in a way that defeats even the fastest supercomputer. Simulating how molecules interact, for instance, means tracking quantum behaviour that grows exponentially with the number of particles. Add a few atoms and the number of states to track explodes past what any classical machine can hold, let alone compute. Richard Feynman put his finger on this in 1981, in a talk arguing that if you want to simulate nature, and nature is quantum, then you had better build a computer that is quantum too. That single insight is the seed the whole field grew from.</p> <h2 id="qubits-superposition-and-entanglement">Qubits, Superposition and Entanglement</h2> <p>The unit of a quantum computer is the qubit. Like a bit it can be 0 or 1, but before you measure it, it can exist in a <strong>superposition</strong> of both at once — not &ldquo;we are unsure which&rdquo;, but genuinely both, in a specific weighted combination. Two qubits can represent four combinations simultaneously, three can represent eight, and the number doubles with each qubit added. This is where the fabled parallelism comes from.</p> <p>The second strange ingredient is <strong>entanglement</strong>. Two qubits can be linked so that their states are correlated no matter how you measure them; act on one and you have, in effect, acted on both. Entanglement is what lets a quantum computer coordinate its qubits into a single coherent calculation rather than a heap of independent coin flips.</p> <p>There is a catch, and it is the whole difficulty of the field. The moment you measure a qubit, the superposition collapses to a single definite 0 or 1. You cannot simply read out all those parallel possibilities. The entire art of quantum algorithm design is arranging the computation so that, through interference, the wrong answers cancel out and the right one is overwhelmingly likely to be the one you measure. It is less like reading a result and more like tuning a system so the answer rings out clearly above the noise.</p> <h2 id="what-a-qubit-is-physically-made-of">What a Qubit Is Physically Made Of</h2><div class="ad-unit ad-in-article" aria-label="Advertisement"> <span class="ad-label">Advertisement</span> <ins class="adsbygoogle" style="display:block;text-align:center" data-ad-client="ca-pub-3726833845844946" data-ad-slot="3291553914" data-ad-format="auto" data-full-width-responsive="true"></ins> <script>(adsbygoogle = window.adsbygoogle || []).push({});</script> </div> <p>One thing the coverage rarely explains is that &ldquo;qubit&rdquo; is an abstraction, and there are several very different ways to build one. This matters, because the engineering headaches — and the hype — depend heavily on which approach a given machine uses.</p> <p>The most prominent today are <strong>superconducting</strong> qubits: tiny loops of superconducting circuit cooled to a hair above absolute zero, where quantum effects dominate. Google&rsquo;s and IBM&rsquo;s headline machines work this way. They are fast but fabulously fragile and demand elaborate refrigeration. A second family, <strong>trapped-ion</strong> qubits, holds individual charged atoms in place with electromagnetic fields and manipulates them with lasers; these tend to be more stable and longer-lived but slower to operate. Others chase <strong>photonic</strong> qubits made of particles of light, or <strong>topological</strong> qubits that would encode information in a way inherently resistant to noise — promising in theory, still largely unbuilt in practice.</p> <p>There is no consensus winner yet, and that alone should temper any confident prediction about timelines. When an entire field cannot agree on what its basic component should be made of, you are early. It is a bit like the messy, competing standards of any young technology before one approach quietly wins — the sort of ferment that eventually settles into the tidy tools we take for granted, the way version control eventually settled into the neat mechanics behind <a href="/story/git-internals-what-happens-when-you-type-git-commit/">what happens when you type a commit</a>. Quantum hardware is nowhere near that settled state.</p> <h2 id="the-milestones-that-actually-mattered">The Milestones That Actually Mattered</h2> <p>The theory came before the hardware by a long way, and a handful of results define the field.</p> <p>In 1985, David Deutsch at Oxford formalised the idea of a universal quantum computer, giving the field a proper theoretical foundation. For a while it was a beautiful curiosity with no killer application.</p> <p>Then came 1994, and Peter Shor at Bell Labs produced the result that made governments pay attention: an algorithm that could factor large numbers exponentially faster than any known classical method. Factoring large numbers is exactly what the RSA encryption protecting much of the internet relies on being hard. Shor&rsquo;s algorithm did not break anything on day one — no machine could run it at scale — but it proved that a sufficiently large quantum computer eventually could.</p> <p>In 1996, Lov Grover, also at Bell Labs, found an algorithm giving a quadratic speed-up for searching unstructured data. Less dramatic than Shor&rsquo;s, but broadly applicable.</p> <p>And in 2019, Google&rsquo;s Sycamore processor, with 54 superconducting qubits, completed a carefully chosen sampling task in about 200 seconds that they estimated would take the best classical supercomputer thousands of years. They called it &ldquo;quantum supremacy&rdquo;. IBM promptly disputed the classical estimate, arguing a supercomputer could do it in days — a useful reminder that these milestones are contested, narrow, and often more symbolic than practical.</p> <h2 id="where-the-hype-runs-ahead-of-reality">Where the Hype Runs Ahead of Reality</h2> <p>Here is the honest state of play. Today&rsquo;s quantum computers are <strong>noisy</strong>. Qubits are absurdly fragile — a stray vibration, a flicker of heat, an errant electromagnetic whisper, and the delicate quantum state decoheres and the calculation is ruined. Keeping qubits stable means cooling them to a hair above absolute zero and shielding them obsessively, and even then they hold their state for only fractions of a second.</p> <p>This is why raw qubit counts are misleading. A headline &ldquo;500-qubit machine&rdquo; tells you little if those qubits are too noisy to run a long calculation. The real prize is <strong>error correction</strong>: spreading one reliable &ldquo;logical&rdquo; qubit across many physical ones so the system can detect and fix errors as it goes. That likely takes thousands of physical qubits per logical one, which is why serious, error-corrected machines are widely expected to be years away rather than around the corner. The physics is sound; the engineering is brutal.</p> <p>So no, quantum computers will not replace your laptop, will not speed up your spreadsheet, and are not about to break every password overnight. For the vast majority of computing they offer no advantage whatsoever. Their power is narrow and specific.</p> <h2 id="the-one-threat-worth-taking-seriously">The One Threat Worth Taking Seriously</h2> <p>The exception that deserves real attention is cryptography. If and when a large, error-corrected quantum computer arrives, Shor&rsquo;s algorithm could break the RSA and elliptic-curve encryption underpinning secure communication today. That is not science fiction; it is a known consequence of a known algorithm, waiting only on hardware.</p> <p>The unsettling part is the &ldquo;harvest now, decrypt later&rdquo; problem: an adversary can record encrypted traffic today and simply store it until a capable machine exists to crack it. Anything meant to stay secret for a decade or more is arguably already at risk. This is why cryptographers have spent years developing <strong>post-quantum cryptography</strong> — new algorithms believed to resist quantum attack — and standards bodies have begun rolling them out. Keeping track of which algorithms and libraries your own systems depend on is exactly the kind of hygiene that a <a href="/story/sbom-software-bill-of-materials-and-why-you-should-care-about-your-dependencies/">software bill of materials makes tractable</a>; you cannot migrate away from vulnerable cryptography if you do not know where it lives in your stack.</p> <p>It is worth being precise about the scope of the threat, because it is narrower than the headlines suggest. Not all encryption is equally exposed. The public-key algorithms used to exchange keys and sign certificates — RSA and elliptic-curve — are the ones Shor&rsquo;s algorithm targets, and those are the priority for migration. The symmetric encryption that protects the bulk of your data, such as the widely used AES, is far more resilient; Grover&rsquo;s algorithm only halves its effective strength, which you can offset simply by doubling the key length. So the picture is not &ldquo;all encryption falls&rdquo; but &ldquo;one specific and very widely used layer needs replacing, on a timeline nobody can pin down precisely&rdquo;. That is serious enough to act on if you hold long-lived secrets, and calm enough that panic is unwarranted.</p> <h2 id="troubleshooting-your-own-expectations">Troubleshooting Your Own Expectations</h2> <p>If you want to engage with quantum computing sensibly rather than getting whipsawed by headlines, a few habits help enormously.</p> <p>When you read a breakthrough claim, ask <strong>how many qubits, and how noisy?</strong> A large machine of poor-quality qubits may do less useful work than a small machine of good ones. Ask <strong>what specific problem</strong> it solved — quantum advantage is always narrow, and a win on a contrived sampling task is not a win on anything you care about. And treat &ldquo;quantum supremacy&rdquo; announcements as scientifically interesting but commercially distant; the gap between a lab demonstration and a dependable product is measured in years and often in whole missing technologies like error correction.</p> <p>Above all, resist the instinct to treat quantum as magic. It is a real, constrained tool that is extraordinary at a handful of things and irrelevant to most. The mistake is the same one people make with any hyped technology, including <a href="/story/what-is-agentic-ai-and-why-is-everyone-talking-about-it/">the current wave of agentic AI</a>: assuming that because something is impressive in a demo, it is imminent and universal. It is usually neither.</p> <h2 id="is-it-worth-your-attention-a-verdict">Is It Worth Your Attention? A Verdict</h2> <p>For almost everyone, the practical answer today is: watch, do not act. You do not need a quantum computer, you cannot usefully buy time on one for your problems yet, and nothing you run daily will be quantum for the foreseeable future. The technology is genuinely years from broad usefulness, and anyone telling you otherwise is selling something.</p> <p>There are two exceptions worth caring about now. If you work in cryptography, security, or anything with a long confidentiality horizon, the migration to post-quantum algorithms is a real, present task, not a future one. And if you simply find the ideas fascinating — the physics really is some of the strangest and most beautiful in all of science — then dig in, because understanding how these machines work is rewarding whether or not you ever touch one. Just keep the wonder and the hype in separate boxes. The wonder is earned; the hype, mostly, is not.</p>