Jerry Chow Applies Quantum Acceleration to the Fusion Fuel Problem
By Ben Lee | 16 Jul, 2026

Jerry Chow is using IBM's quantum computer to accelerate the process of optimizing the molecular configuration for a FLiBe molten salt blanket to breed and recover tritium fuel to power fusion reactors.

IBM's Jerry M. Chow knows that quantum computers aren't yet ready for prime time on their own.  But he's taking the lead in melding IBM’s quantum machines into a hybrid scientific workflow that combines AI and supercomputers for the fusion-blanket project with Oak Ridge National Laboratory and Cleveland Clinic.

The project uses quantum-centric supercomputing to calculate nine molecular configurations of FLiBe, a molten salt made from fluorine, lithium and beryllium considered a leading candidate for breeding and recovering tritium in fusion reactors. 

The fusion problem is this: many proposed fusion reactors need tritium, but tritium is scarce. One way to make it is to wrap the reactor in a blanket material containing lithium. Neutrons from the fusion reaction strike the blanket, producing tritium. FLiBe is attractive because it can act as a coolant, shielding material and tritium-breeding medium. But under neutron radiation, heat and magnetic fields, its chemistry keeps changing, and researchers need to know whether tritium will be easily recoverable or will bind into difficult chemical forms. IBM describes that as one of the hardest parts of the engineering problem: predicting FLiBe’s energetics, stability and interaction with tritium at the quantum-mechanical level. 

What Chow emphasizes is the workflow. In his own public summary, he says the problem is modeling how tritium binds in FLiBe, and that the team split the work because the electronic interactions driving tritium’s behavior are strongly correlated and hard for classical methods to capture accurately. The surrounding molten-salt environment remains on classical high-performance computers, while the chemically active fragment is isolated and sent to IBM quantum systems. 

The technique is called an embedded wavefunction method. In plain English, the researchers don’t try to make today’s quantum computer simulate the entire molten-salt blanket. They cut the problem into fragments. Classical computers handle the easier or larger environmental parts, while the quantum processor handles the small but chemically delicate fragment where electron correlation matters most. That is the part where quantum computing may have an edge, because electrons themselves obey quantum mechanics. Chow says the IBM systems used extended sample-based quantum diagonalization, or ExtSQD, in which the quantum processor produces correlated samples and classical computers perform the diagonalization step. 

The immediate scientific target was tritium binding energy: how strongly tritium attaches to different local structures in FLiBe. That matters because tritium’s chemical form determines whether it can be extracted efficiently. If tritium gets trapped in stubborn or corrosive forms, the blanket may fail as a practical fuel-production system. IBM says the hybrid calculations let scientists identify what configurations the atoms move through and extract hidden properties, including how strongly and by what mechanism each configuration binds tritium. 

The arXiv paper reports the first application of heterogeneous quantum-classical computing to tritium binding in FLiBe. It says the researchers took clusters from ab initio molecular dynamics, partitioned them into atom-centered fragments with embedded-wavefunction methods, and solved the largest fragments on IBM quantum hardware using ExtSQD. Across nine clusters, the workflow reproduced fragment ground-state energies close to full configuration interaction benchmarks, with mean absolute deviation of 0.3 kcal/mol. 

Just as important, the work exposed where the current method still falls short. The paper says the biggest source of error wasn’t the quantum solution of the fragment; it was how the fragments were constructed. That is valuable because it tells IBM and its partners where to improve the workflow before scaling to larger, more realistic molten-salt systems. 

Jerry M. Chow is one of IBM’s central builders of superconducting quantum computing, especially the move from lab experiments to deployable quantum systems.

Chow is an IBM Fellow and is listed as Director of Quantum Systems & Runtime Technology at IBM Quantum; his public profile describes his expertise as the design, measurement and integration of superconducting qubits. He earned his Ph.D. at Yale in 2010, where his dissertation was on quantum information processing with superconducting qubits. 

His early IBM work helped establish the hardware foundations for IBM’s quantum program. IBM’s 2022 Fellow profile says Chow led IBM quantum hardware toward cloud deployment, and credits his work on two-qubit gates, large-system calibration, stability and integration with helping move quantum computing from research demonstrations into systems development. 

Technically, Chow’s specialty has been making superconducting qubits behave well enough to scale. He co-authored important early work on high-fidelity universal gate sets in superconducting qubits and on implementing pieces of surface-code-style error correction, both central to the long-term goal of fault-tolerant quantum computers. 

In IBM’s more recent quantum push, Chow has become associated with quantum-centric supercomputing: the idea that quantum processors should be integrated with CPUs, GPUs and classical high-performance computing rather than treated as standalone machines. A 2026 reference-architecture paper co-authored by Chow describes quantum processors as specialized accelerators inside hybrid HPC systems, especially for chemistry and materials workloads. 

That framework helps explain his role in the fusion-blanket materials work. Chow’s contribution is not mainly as a fusion-materials scientist, but as a quantum-systems architect helping IBM use today’s quantum hardware on the hardest quantum-chemistry fragments inside a larger classical workflow. This is the same IBM strategy that underlies its broader roadmap toward fault-tolerant quantum computing by 2029, which Chow also co-authored. 

A concise way to describe him in an article would be:

Jerry M. Chow, an IBM Fellow and one of the company’s leading superconducting-qubit engineers, has helped turn IBM Quantum from a laboratory research effort into a cloud-accessible quantum-computing platform. His work on two-qubit gates, calibration, system stability and modular quantum architecture has made him a key figure in IBM’s shift toward quantum-centric supercomputing, in which quantum processors work alongside classical supercomputers to tackle hard chemistry, materials and physics problems.