IBM, NVIDIA and Fermilab Build Quantum-Classical Computing Platform

November 16, 2026
IBM, NVIDIA and Fermilab Build Quantum-Classical Computing Platform
  • IBM, NVIDIA and Fermilab plan to build a modular quantum-computing co-design platform pairing NVIDIA DGX B200 systems with IBM quantum computers to investigate how quantum and classical processors can scale together.

  • The platform will support research into quantum error-correction codes, decoding algorithms and quantum-classical applications, drawing on Fermilab’s expertise in high-energy and nuclear physics and quantum error correction.

  • Researchers will first model communication between quantum modules, measuring link errors, latency and feedback to guide the development of physical quantum interconnects.

  • In a later phase, Fermilab’s SQMS Center plans to integrate ultracoherent superconducting cavities and study their potential use as quantum memories, entanglement buffers and tools for bosonic error correction.

  • Fermilab chief technology officer and SQMS director Anna Grassellino said the initiative will help build capabilities and a research community for a potential future Department of Energy quantum-computing user facility.

  • The work, supported by Japan’s NEDO, was named a finalist for the 2026 ACM Gordon Bell Prize.

  • NVIDIA is expanding the open-source Slurm workload manager to schedule quantum processors alongside CPUs and GPUs, with IBM contributing integration through its Quantum Resource Management Interface.

  • The vendor-neutral interface, developed by IBM, NVIDIA and Pasqal with the open-source community, lets Slurm manage shared quantum-processing resources—including reservations and time allocation—within familiar high-performance-computing workflows.

  • Separately, NVIDIA released its TCBP-JOBS decoding algorithm through the CUDA-Q QEC 0.9.0 library, reporting a maximum 13,000-fold speedup on specified benchmarks against unsparsified long-beam Tesseract—not across all workloads.

  • In one research computation, the decoder cut processing time from 12 weeks to 45 minutes.

  • In research by IBM, RIKEN and NVIDIA, a quantum-classical workflow simulated the Fe4S4 iron-sulfur cluster using circuits of up to 4,000 two-qubit operations on 72 qubits and collecting five million samples.

  • In separate chemistry research, IBM and RIKEN used an IBM quantum computer with RIKEN’s NVIDIA GB200-powered ROQUO supercomputer to calculate binding energies for protein-drug pairs, modeling complexes of up to 12,635 atoms compared with a previous 303-atom benchmark.

Summary based on 7 sources


Get a daily email with more Tech stories

More Stories