← Back to blog
5 min read News

IBM, RIKEN, and Cleveland Clinic cross the 12,000-atom barrier

IBM, RIKEN, and Cleveland Clinic say they modeled protein complexes up to 12,635 atoms with quantum-classical workflows, a real scaling result for quantum chemistry rather than a qubit-count headline.

newsIBMRIKENCleveland Clinicquantum-chemistry

IBM, RIKEN, and Cleveland Clinic say they have used quantum computers and classical supercomputers to simulate protein complexes spanning up to 12,635 atoms. The claim matters because it is specific, bounded, and tied to a measurable scientific workflow: quantum chemistry on biologically relevant molecules.

The work is described in IBM’s newsroom post and a matching RIKEN announcement, with the technical details posted on arXiv. IBM says the team used its 156-qubit Heron processors at Cleveland Clinic and RIKEN, along with supercomputers Fugaku and Miyabi-G, to run a heterogeneous quantum-classical workflow on protein-ligand systems.

That is a narrower and more useful story than the usual quantum headline. It does not claim quantum advantage in the broad sense. It does not claim fault tolerance. It does show that the field keeps pushing the scale of problems that quantum hardware can touch when it is paired with classical infrastructure.

What the result actually shows

The strongest number here is not the atom count by itself. It is the combination of scale, resource use, and improvement over prior work.

According to the arXiv paper, the team sampled fragment electronic configurations on two 156-qubit processors, used up to 94 qubits in parts of the workflow, ran 9,200 circuits for more than 100 hours, and collected 1.3 billion measurement outcomes. The paper says the result covers two protein-ligand complexes of 11,608 and 12,635 atoms, with a 40-fold increase in system size and up to a 210x improvement in accuracy over previous results.

Those are the details that make the story credible. They are concrete, reproducible, and tied to a named workflow: quantum embedding plus heterogeneous quantum-classical supercomputing.

Why this matters for quantum computing

Drug discovery has always been one of the field’s most plausible applications, but it has also been one of the hardest to prove in practice. Chemistry problems grow quickly, and classical methods get expensive fast. The IBM-led result is important because it moves the discussion from vague promise toward a more operational question: how large a biomolecular problem can be handled when quantum processors are used as part of a larger stack?

That does not mean the hardware is ready to replace classical computation. It means the hybrid approach can now be tested on larger systems with better accuracy than before. The result also gives the field a more honest benchmark than qubit totals alone. A 156-qubit device is interesting. A 12,635-atom simulation is harder to dismiss.

Still, the limits matter. The work depends on a specialized workflow, major classical compute resources, and careful decomposition of the molecular problem. It is a milestone in quantum chemistry, not proof that general-purpose quantum computers are around the corner.

The useful takeaway

This is one of the better quantum stories of the month because the numbers line up with the claim. The press release, the blog post, and the preprint all point to the same thing: quantum hardware is becoming more useful as part of a hybrid scientific workflow, especially for chemistry problems that are too large for the earlier generation of demonstrations.

If the next round of progress keeps improving scale and accuracy while reducing the amount of manual orchestration required, quantum chemistry will start to look less like a promise and more like an engineering pipeline.

Sources & Further Reading

Primary sources:

Context & analysis: