Quantum computers are proving they can contribute meaningful results
At the intersection of quantum mechanics and molecular biology, three institutions have quietly moved a boundary that once seemed fixed. IBM, Cleveland Clinic, and RIKEN have simulated a protein complex of 12,635 atoms — the largest ever modeled on quantum hardware — using a hybrid architecture that pairs quantum processors with classical supercomputers. The achievement matters not as a technical trophy but as a signal: the long-promised practical utility of quantum computing is beginning to arrive, and its first meaningful address may be the pharmacy.
- Drug discovery has long been held hostage by a computational wall — accurately modeling how drug candidates bind to proteins becomes exponentially harder as molecules grow, costing years and hundreds of millions of dollars per failed attempt.
- In six months, the same hybrid method went from handling molecules 40 times smaller to simulating a 12,635-atom complex, with accuracy on a key workflow step improving by up to 210 times — a compression of progress that is difficult to overstate.
- The breakthrough required assembling a system of rare scale: IBM's 156-qubit Quantum Heron processors at two sites, orchestrated by Fugaku and Miyabi-G — two of the world's most powerful classical supercomputers — running nearly 6,000 quantum operations across up to 94 qubits.
- A newly developed hybrid algorithm called EWF-TrimSQD was central to the leap, slashing computational overhead and allowing quantum hardware to directly represent the chemistry of biologically meaningful molecules for the first time.
- The field is watching a quiet shift in how quantum computing measures itself — no longer by qubits and error rates alone, but by the size and significance of real problems it can now help solve.
Three institutions — Cleveland Clinic, RIKEN, and IBM — have crossed a threshold researchers have been chasing for years. Together, they simulated a protein complex containing 12,635 atoms, the largest biological molecule ever modeled on quantum hardware. The achievement signals something the field has been waiting to hear: quantum computers are becoming practical tools for real scientific problems.
The stakes are rooted in a bottleneck that has plagued drug discovery for decades. Determining whether a drug candidate will bind to a protein target — a foundational question in medicine development — involves calculations so complex that classical methods break down as molecules grow larger. Getting this right early could compress timelines that currently stretch over a decade. The team set out to test whether quantum computers might crack what classical machines have found so difficult.
What they built was a hybrid system. Classical computers divided protein-ligand complexes into manageable fragments. IBM's 156-qubit Quantum Heron processors — operating at Cleveland Clinic in Ohio and at RIKEN in Japan — then calculated the quantum-mechanical behavior of those pieces, with results reassembled on classical machines. The supercomputers Fugaku and Miyabi-G handled the orchestration. In certain stretches, the quantum processors ran nearly 6,000 operations across up to 94 qubits.
The scale of improvement is striking. Six months earlier, the same method could handle molecules roughly 40 times smaller. A new hybrid algorithm called EWF-TrimSQD dramatically reduced computational overhead and made it possible to represent molecular chemistry directly on quantum hardware — a key enabler of the leap.
Kenneth Merz of Cleveland Clinic described crossing the 12,000-atom barrier as expanding what quantum computing could do with biologically meaningful molecules. IBM's Jay Gambetta noted the broader shift: for years, quantum computing has been a promise. Now it is producing results that matter to biologists and chemists working in actual laboratories.
This milestone is part of a clear trajectory. Earlier work modeled electronic states in iron sulfides; more recently, the team completed the first full quantum-centric simulation of Trp-cage, a 303-atom protein. Each step has been incremental, but the direction is unmistakable. If quantum systems can scale further while maintaining accuracy, they may eventually help predict drug-protein interactions and simulate enzyme mechanisms that today require physical experimentation — compressing not just timelines, but the distance between discovery and cure.
Three institutions—Cleveland Clinic, RIKEN, and IBM—have crossed a threshold that researchers have been chasing for years. Using quantum computers paired with some of the world's most powerful classical supercomputers, they simulated a protein complex containing 12,635 atoms. It is the largest biological molecule ever modeled on quantum hardware, and it signals something the field has been waiting to hear: quantum computers are becoming practical tools for real scientific problems.
The work matters because of a bottleneck that has plagued drug discovery for decades. When researchers want to know whether a drug candidate will bind to a protein target—a fundamental question in developing any new medicine—they face a calculation so complex that existing computational methods struggle to solve it accurately as molecules grow larger. Getting this right early could compress drug development timelines that currently stretch over a decade and consume hundreds of millions of dollars. The team decided to ask whether quantum computers might crack what classical machines have found so difficult.
What they built was a hybrid system. Classical computers broke the protein-ligand complexes into manageable pieces. IBM's 156-qubit Quantum Heron processors—running at Cleveland Clinic in Ohio and at RIKEN in Japan—then calculated the quantum-mechanical behavior of those fragments. The results were reassembled on classical machines to reconstruct the full molecular picture. The two most powerful classical supercomputers in the world, Fugaku at RIKEN and Miyabi-G operated by Japanese universities, handled the orchestration. In certain parts of the simulation, the quantum processors ran nearly 6,000 quantum operations across up to 94 qubits.
The scale of improvement is striking. Six months earlier, this same method could handle molecules roughly 40 times smaller. In that same period, the accuracy of a key step in the workflow improved by up to 210 times. The leap came from both algorithmic innovation and raw computing power. The team developed a new hybrid algorithm called EWF-TrimSQD that dramatically reduced computational overhead and made it possible to represent the chemistry of these systems directly on quantum hardware.
Kenneth Merz, the lead researcher at Cleveland Clinic's Computational Life Sciences Department, framed the moment plainly: crossing the 12,000-atom barrier expanded the scale of what quantum computing could do with biologically meaningful molecules. Jay Gambetta, IBM's Director of Research, offered a broader observation. For years, quantum computing has been a promise. Now it is producing results that matter. The systems they simulated are the kind of molecules that biologists and chemists actually work with in laboratories.
This is not the first milestone from these three institutions. They published earlier work on modeling electronic states in molecules, demonstrated on iron sulfides and featured on the cover of Science Advances. More recently, they completed the first full quantum-centric simulation of a 303-atom molecule called Trp-cage, a protein made of 20 amino acids. Each step has been incremental, but the trajectory is clear.
The researchers see this as a beginning. If quantum computers can scale to larger molecular systems while maintaining accuracy, they could help predict how medicines interact with protein targets—one of the two fundamental capabilities that computational drug discovery depends on. The other is accurately computing molecular energies, which this work demonstrates quantum-centric supercomputing can support. Looking further ahead, the same approach might eventually enable simulation of enzyme catalysts and drug mechanisms that today can only be studied through physical experimentation.
For most of its history, quantum computing has measured itself by qubits, gates, and error rates—the machinery of the field. This milestone marks a shift. Progress now means something different: the size and significance of the real problems quantum computers can help solve. That shift from promise to practice is what makes this moment worth watching.
Citas Notables
By crossing the 12,000-atom barrier, we have significantly expanded the scale of biologically meaningful molecular simulations possible with quantum computing.— Kenneth Merz, Cleveland Clinic's Computational Life Sciences Department
Quantum computers are no longer proving they are viable tools—they are proving they can contribute meaningful results in quantum-centric supercomputing architectures.— Jay Gambetta, IBM Director of Research
La Conversación del Hearth Otra perspectiva de la historia
Why does simulating a protein matter so much to drug discovery? Can't researchers just test drug candidates in a lab?
They can, but it's slow and expensive. Before you test anything in a lab, you want to know computationally whether a drug will even bind to its target protein. That calculation—predicting the interaction—is one of the hardest problems in computational chemistry, especially as molecules get larger.
And quantum computers are better at this than classical computers?
Not better at everything. But for certain kinds of calculations—the quantum-mechanical behavior of atoms and electrons—quantum processors have a natural advantage. The trick is knowing which parts of the problem to give to quantum and which to give to classical machines.
So they split the work?
Exactly. Classical computers break the protein into pieces. Quantum computers calculate the quantum behavior of those pieces. Then classical computers reassemble the answer. Each tool does what it does best.
How much faster does this make drug discovery?
That's the hope, but we're not there yet. This is proof that quantum computers can handle molecules large enough to be biologically relevant. If accuracy scales with size, then yes, you could predict drug interactions faster and earlier in development. That could shorten timelines that currently take a decade.
What was the biggest technical hurdle they overcame?
The algorithm. They developed something called EWF-TrimSQD that reduced the computational overhead dramatically. It let them represent the chemistry directly on quantum hardware instead of having to translate everything into a form quantum computers could handle.
Is this the end of the road, or is there more to come?
Much more. They simulated a protein with 12,635 atoms. That's large, but real drug targets can be bigger. The path forward is clear—keep scaling, keep improving accuracy, and eventually integrate this into actual drug discovery workflows.