DALLAS - IBM, NVIDIA, and Fermilab announced on Nov. 16, 2026, at SC26 a joint plan to build a modular quantum-classical computing platform in Illinois.
Co-Locating Quantum and AI Hardware
The partnership will place NVIDIA DGX B200 systems directly alongside IBM quantum computers at Fermilab's Superconducting Quantum Materials and Systems Center. The deployment includes IBM Quantum Nighthawk processors. These can execute up to 7,500 two-qubit operations across 120 qubits. Sampling rates reach 100,000 shots per second.
Connecting quantum processors directly with AI supercomputers allows researchers to offload heavy error mitigation and pre-processing tasks in real time.
Molecular Simulation Gains and Supercomputing Integration
In a parallel demonstration with Japan's RIKEN, IBM and NVIDIA simulated an iron-sulfur cluster. They ran 4,000 two-qubit operations on 72 qubits. Collecting five million samples delivered a 79-fold accuracy improvement. Energy estimates came within 4.3 millihartree of classical benchmarks.
Separately, the Slurm 26.11 workload manager now uses a standardized interface. It integrates quantum processors across nearly 30 supercomputing centers. Those centers deploy over 50 quantum machines.
Co-Designing Fault-Tolerant Science Tools
Fermilab CTO Anna Grassellino emphasized that bringing high-energy and nuclear physicists together is critical. That collaboration is needed to co-design practical applications and modular architectures. Joint engineering teams will test algorithms and hardware interfaces at the Illinois facility. The goal is to make fault-tolerant quantum computing viable for basic physics research.



