NVIDIA Announces Collaboration With IBM and Fermi National Accelerator Laboratory to Launch Quantum-Classical Codesign Platform, Plus Error-Correction Breakthrough to Advance Practical Quantum Computing
AI-assisted digest, checked against the original source before publishing. How we work
NVIDIA, IBM and Fermilab are building a quantum-classical codesign platform, while a new CUDA-Q decoder claims up to a 13,000x speedup for error-correction research.
NVIDIA announced a new set of quantum-computing collaborations with IBM and the U.S. Department of Energy's Fermi National Accelerator Laboratory, covering next-generation modular quantum architectures, scientific applications and integration of quantum processors into supercomputing workflows. The announcement pairs a jointly developed codesign platform with a separate algorithmic result: a new decoder for quantum error correction that NVIDIA says is up to 13,000x faster than the leading existing algorithms.
The hardware side involves colocating NVIDIA DGX B200 systems with IBM facility quantum computers, including machines powered by IBM Quantum Nighthawk, which NVIDIA describes as the fastest quantum processor IBM has delivered. That combined system underpins the IBM, Fermilab and NVIDIA modular quantum codesign platform, which is intended to let researchers develop and evaluate error-correction codes, decoding algorithms and quantum-classical applications aimed at discoveries in high-energy and nuclear physics. The platform is also meant to test how classical systems can scale alongside multiple linked quantum processors within modular architectures.
NVIDIA is also expanding Slurm to let quantum processors be scheduled inside supercomputing workflows, with IBM contributing integration through its Quantum Resource Management Interface. The move frames quantum hardware as another schedulable resource alongside GPUs in existing HPC job-management tooling rather than a separate operational silo.
What to watch next is whether the codesign platform produces published results on real IBM hardware, and how much of the 13,000x decoding advantage survives outside the benchmark conditions NVIDIA used. Also open: how far DGX B200–quantum colocation extends beyond IBM's own facilities, and whether Slurm-based QPU scheduling becomes standard practice or remains a research integration.
For builders, the near-term signal is that error-correction tooling is moving into the same accelerated-computing stack used for AI workloads, which lowers the cost of experimenting with decoder designs even before fault-tolerant hardware arrives.
Key facts
- • NVIDIA's new TCBP-JOBS decoder ships in the CUDA-Q platform.
- • Reported decoding speedup reaches up to 13,000x versus leading algorithms.
- • Decoder research time drops from 12 weeks to 45 minutes.
- • DGX B200 systems colocated with IBM Quantum Nighthawk machines.
- • IBM contributes its Quantum Resource Management Interface to Slurm.
Why it matters
Decoder speed was a practical bottleneck in exploring error-correction designs, so a 13,000x gain compresses iteration cycles from months to under an hour for research teams. Builders get quantum processors exposed as schedulable resources through Slurm, meaning GPU-cluster operators can begin wiring QPUs into existing workflows without new job-management tooling. The open question is whether thes
Spotted an error? Send a correction.
Get the day's AI news in one email
The Daily Epoch · 60-second read · unsubscribe anytime