IBM, Cleveland Clinic, and RIKEN Simulate a 12,635-Atom Protein, Setting a Quantum-Centric Computing Record

A joint team from IBM, Cleveland Clinic, and Japan's RIKEN has performed what they describe as the largest biologically meaningful molecular simulation ever run on quantum hardware: a full electronic-structure calculation of the protein trypsin at 12,635 atoms, alongside a sister run on T4-lysozyme at 11,608 atoms, both modeled in liquid water. The announcement, published May 5, marks the field's first quantitative step from quantum chemistry toy systems into protein-scale biology.

The compute stack is the headline. The team ran quantum sampling on two of IBM's 156-qubit Quantum Heron r2 processors, using up to 94 qubits for the active subspace where electron correlation matters most, and farmed the resulting amplitudes to two of the world's largest classical supercomputers — Japan's Fugaku and the new Miyabi-G — for downstream processing. The full pipeline executed 9,200 quantum circuits across more than 100 hours of runtime, collecting roughly 1.3 billion measurements. The combined heterogeneous quantum-classical (HQC) calculation is the largest electronic-structure run ever attempted on real hardware.

The deltas against the field's prior state of the art are what justify the breathless framing. The result represents a 40-fold increase in simulated system size and a 210x improvement in accuracy over comparable simulations published just six months earlier. Those gains do not come from raw qubit count alone — they come from a new sample-based quantum diagonalization algorithm and a tighter QPU/CPU/GPU choreography that lets the classical side absorb noise the quantum side would otherwise drown in.

For drug discovery, the practical implication is concrete. Trypsin is a workhorse enzyme used across diagnostics and as a model for serine-protease drug targets; lysozyme is a standard benchmark for protein-ligand binding studies. Being able to compute their electronic structure quantum-mechanically — rather than approximating with density functional theory or molecular mechanics — opens the door to first-principles calculation of binding affinities for targets that have so far been refractory to classical simulation. Cleveland Clinic's interest is unsubtle: it runs one of the most productive translational pipelines in American medicine and has been searching for a way to feed quantum chemistry results directly into structure-based drug design.

The work also moves the goalposts on what counts as a "useful" quantum advantage. The team is not claiming supremacy over classical methods on this particular problem — Fugaku and Miyabi-G did most of the arithmetic. They are claiming that a hybrid architecture can model biological systems at a scale classical methods alone cannot reach accurately, and at error bars tight enough for chemistry, not headlines. That distinction is what separates demo physics from drug discovery, and on May 5 it crossed a threshold the field has been chasing for a decade.