Quantum Computers Could One Day Help Chemists Check the Molecules Behind New Medicines

When a chemist makes a new molecule, maybe a candidate medicine, one of the first questions is simple: did I make what I meant to make? A workhorse for answering it is NMR, a technique that reads the magnetic chatter of atomic nuclei. Quantinuum’s write-up of the work puts it plainly: in pharmaceutical research, NMR helps identify and characterize drug candidates. Matching a measured spectrum to a predicted one often takes heavy computing.

In a preprint posted September 15, 2026, researchers from HQS Quantum Simulations in Karlsruhe and from Quantinuum used Quantinuum’s H2 trapped-ion quantum computer to simulate the proton NMR spectrum of a molecule called 1,2-di-tert-butyl-diphosphane. It is a small, awkward molecule: two phosphorus and twenty hydrogen nuclei, which the paper calls a classically challenging benchmark molecule.

The quantum machine reproduced the main features of the classical reference spectrum, including a double-peak structure near 4.1 ppm that an earlier 22-qubit demonstration on other hardware had failed to reproduce.

Three things to know

• It was a big run for today’s machines. The team boiled the 22 nuclear spins down to an effective 21-spin model and ran it on 42 qubits, 21 for the molecule and 21 helpers. The deepest circuits reached a two-qubit gate depth of 1,442 across 70 time steps, about 29 milliseconds of simulated magnetic evolution.
• It is not faster than a regular computer. The authors do not claim quantum advantage. Quantinuum’s post says exact classical simulations are typically limited to around 25 interacting spins, and approximation methods often reach roughly 40 to 50. By that measure, this molecule sits within reach of classical methods, which is exactly why it makes a good test: there is a trusted classical answer to check against. The paper goes further, saying classical methods appear to cover a large portion of practically relevant use cases for ordinary high-field liquid NMR.
• It is not yet sharp enough to use. The resolution was about 0.07 ppm, too coarse for the roughly 0.01 ppm fine structure chemists care about. The quantum spectrum also misses two sharp peaks that the simplified model left out. According to Quantinuum, its collaborators at HQS, who co-wrote the paper and develop one of the leading classical NMR simulation packages, say the resolution is still insufficient for routine use by practicing spectroscopists.

So why does it matter

Because this is the kind of milestone that tells you where the road goes. NMR is a natural job for a quantum computer: the molecule’s nuclei are themselves a quantum system, and each one can be mapped onto a qubit. Getting a deep, noisy machine to produce a spectrum whose main features agree with the classical reference shows the pipeline works from start to finish.

The authors are candid about the gap. Resolving the fine structure would need about 200 milliseconds of simulated time, more than 500 time steps, which means circuits far deeper than present-day hardware can run. The error suppression they used also does not directly correct some kinds of random errors. They point to ultralow-field NMR, a different measuring regime, as a more promising place for quantum machines to pull ahead.

For now, the people this could one day help are the chemists staring at a tangle of peaks late in the evening, trying to confirm a structure before a promising compound moves forward. Nothing on their bench changes today. But the test molecule now has an answer from a second, very different kind of machine, and in its main features, it agrees.