Can a quantum computer help choose the next drug experiment?

Choosing which molecule to test can consume money and months of laboratory work. A new partnership between Quandela and a Korean biotechnology institute puts that decision at the centre of a quantum-computing project: can a calculation help a scientist choose the next experiment? A molecule does not care how fashionable the machine studying it is. It behaves according to physics and chemistry. A promising calculation still has to survive contact with an experiment, and a promising experiment still has to survive many more tests before it becomes a medicine. That is the distance hidden inside the phrase “quantum drug discovery.”

On 7 October 2026, photonic quantum-computing company Quandela announced a biotechnology collaboration with the Korea Research Institute of Bioscience and Biotechnology, known as KRIBB. The agreement was signed in Paris on 8 September. The partners describe work combining quantum computing, artificial intelligence and conventional computation. The announcement is evidence of a research collaboration, not evidence that a quantum computer has discovered an approved medicine or beaten the best alternatives on a useful drug-development task.^[1]

The proposed applications include designing molecules for targeted protein degradation, improving cellular therapies and finding therapeutic targets for rare diseases. The partners describe AI narrowing candidates, conventional computing handling preparation and modelling, and photonic quantum methods exploring difficult combinations. Those are proposed responsibilities, not evidence that each component has already improved a biological outcome.^[1]

Start with the decision, not the device

Consider a hypothetical research team deciding which of several candidate molecules deserves an expensive experiment. The team does not need a machine to announce that chemistry is complicated. It needs information that changes the ordering of its options, improves confidence in a comparison or identifies a mistake in its assumptions. The calculation might concern a small but difficult part of a much larger system. The rest of the workflow could remain entirely conventional. That would not make a quantum contribution less valuable. A specialised instrument earns its place by improving a particular job, not by replacing every instrument in the building.

Now imagine the new calculation returns a number with many decimal places. That looks impressive until the scientist asks whether the uncertainty is small enough to change the decision. A beautifully precise answer to an oversimplified model can be less useful than a rougher answer that describes the relevant environment. More digits are not automatically more knowledge. This is the distinction a serious collaboration should make visible: the difference between computing a property and improving a decision. QubitWire’s interpretation is that the second question is the better test of commercial relevance. It is an editorial framework, not a claim that the new partners have already passed it.

Why three kinds of computing can belong in one project

“Quantum plus AI” can sound like two fashionable labels glued together. It becomes meaningful only when the responsibilities are clear. One component might organise existing information. Another might propose candidate models. A quantum routine might address a narrowly defined calculation. Conventional software would still prepare inputs, manage results and connect the work to the laboratory. The useful question is not whether all three technologies appear in a diagram. It is whether removing one of them makes the result meaningfully worse, more expensive or less reliable. That counterfactual is a practical way to examine a hybrid-computing claim.

An everyday comparison is a hospital laboratory. A microscope, a database and a statistical package can all contribute to one investigation without doing the same thing. Their coexistence is not proof that each is essential to every task. The same discipline should apply when the tools are more exotic. The partners’ announcement provides a reason to follow that division of labour. It does not, by itself, establish the quality of the final workflow. Readers should look for a clearly defined problem, an accessible method and a comparison that makes the contribution of each component understandable.^[1]

The comparison that would actually be persuasive

Suppose two teams are asked to predict a property before an experiment is performed. One uses an established computational approach. The other adds a quantum step. To make the comparison useful, both teams should work toward the same scientific target and report the resources their workflows consume. It would be unfair to give one method a carefully tuned model and compare it with a deliberately weak alternative. It would be equally misleading to report the time spent inside the quantum processor while ignoring preparation, repeated measurements, analysis and failed runs. The scientist waiting for a result experiences the whole process.

There is also a less obvious issue: how much information was available when the prediction was made? A result assessed after researchers know the experimental answer is different from a genuinely prospective prediction. Both can be valuable, but they answer different questions. An honest account should say which occurred. A strong result might even show that quantum computing was not the best option for the first chosen problem. That could save a research programme from chasing an attractive but unsuitable demonstration. Negative results are not a failure of scientific collaboration when they clarify where the tool does and does not belong.

A better result need not arrive as a blockbuster headline

The first useful output could be a reproducible method, a carefully documented limitation or a problem that other groups can test. Those outcomes are less dramatic than “quantum discovers a drug,” but they are easier for the wider research community to use. A reproducible workflow lets a second team ask whether the result survives a different implementation. A clearly stated limitation prevents another group from spending months on a regime the method cannot handle. A well-chosen benchmark can connect hardware developers with a scientific need that is more concrete than an abstract performance score.

This is also why an industrial partnership and a scientific breakthrough should not share the same label. A partnership describes an intention and an organisational relationship. A paper describes a claim and its supporting method. Independent reproduction tests whether that claim survives outside the original setting. These are different milestones, not interchangeable marketing terms. Quandela’s announcement says initial findings have been shared and describes plans for publication and reproducibility. Those plans create a useful follow-up question: what becomes publicly inspectable, and when? They are not a substitute for the material itself.^[1]

What a reader should ask when the results arrive

The first question is wonderfully plain: what did the scientists need to know? A study that cannot explain its scientific question in ordinary language will be difficult for non-specialists to evaluate, even if its mathematics is sophisticated. Next comes the comparison. What was the strongest relevant method without the quantum component? Was it given a fair implementation? What uncertainty remained? A small advantage on an artificial example and a useful advantage in a realistic setting are both worth reporting, but not under the same headline.

Finally, ask what changed in the laboratory or the research plan. Did the result identify a candidate worth testing, rule out an assumption or improve a prediction that could be checked? A computational milestone can matter before it changes a drug-development decision, but the distance between those stages should remain visible. None of these questions requires a reader to dismiss quantum computing. They make enthusiasm more informative. The most persuasive story will be the one that lets a curious outsider explain both the achievement and the remaining uncertainty to somebody else.

Why this matters beyond one partnership

The broader lesson is about how emerging technologies meet real problems. A technology company naturally wants a problem that demonstrates its machine. A domain scientist wants an answer that improves the science. A productive collaboration has to negotiate that difference rather than hide it behind a shared logo. That negotiation can also improve communication. “We used a quantum computer” tells the public about the tool. “We tested whether a difficult calculation changes this experimental choice” tells the public about the purpose. The latter is a much better starting point for understanding progress.

For now, the defensible story is that a quantum-hardware company and a biotechnology institute have agreed to investigate a shared research agenda. The more consequential story will emerge only when the work produces something others can examine. Keeping those statements separate does not make the development boring. It tells readers exactly what to watch next. A quantum computer will not make a medicine valuable by association. Its value will come from helping researchers know something useful that they could not otherwise know as well. That is a demanding test. It is also a far more interesting one than the label on the machine.