IBM, NVIDIA, and Fermilab's Quantum Co-Design Platform Explained
As seen on the 24/7 Wall St. homepage on November 16, 2026.
IBM, Fermilab, and NVIDIA Announce Plans for New Co-Design Platform to Accelerate the Development of Next-Generation, Modular Quantum Computers
DGX B200 systems get co-located with IBM quantum computers at Fermilab, putting GPUs inside the quantum stack. Supporting work with IBM and RIKEN hit a 79-fold accuracy gain simulating an iron-sulfur cluster, and a new open-source layer lets Slurm schedule quantum processors like CPUs and GPUs. Accelerated computing becomes the default quantum-classical substrate.
Continue ReadingShow less
The core of the announcement is a plan to physically co-locate NVIDIA DGX B200 systems with IBM quantum computers at Fermilab, putting GPU horsepower directly inside the quantum computing stack rather than treating classical and quantum resources as separate systems that communicate at arm's length. The host site, Fermilab's Superconducting Quantum Materials and Systems Center, brings expertise in ultracoherent superconducting cavities and quantum memories that neither IBM nor NVIDIA could replicate on their own.
The IBM quantum hardware involved is the Nighthawk processor, which IBM describes as the most powerful quantum processor it has ever built. It runs circuits of up to 7,500 two-qubit operations on 120 qubits at speeds of up to 100,000 samples per second. The intent is to let multiple quantum processors work together at scale, with classical GPU compute running alongside them rather than waiting on results from a distance.
The Slurm integration, developed with IBM, NVIDIA, and Pasqal through an open-source layer called the Quantum Resource Management Interface (QRMI), is arguably the most practically significant piece for existing HPC centers. Slurm is the scheduler behind many of the world's largest supercomputers, but it previously had no native concept of a quantum processor. QRMI plugs into Slurm's Hierarchical Resources scheduling capabilities so that QPU time can be allocated, reserved, and accounted for using the same policies HPC administrators already apply to CPUs and GPUs, with no separate queue or login required.
Sponsored
The scientific results published alongside the announcement come from IBM, RIKEN, and NVIDIA and involve simulating Fe4S4, an iron-sulfur cluster found in living cell proteins. Using a technique called sample-based quantum diagonalization, the team ran circuits of up to 4,000 two-qubit operations on 72 qubits and collected 5 million samples, integrating IBM quantum hardware with NVIDIA GPUs including those inside RIKEN's ROQUO supercomputer. The result was a 79-fold improvement in accuracy over earlier runs of the same experiment.
The final accuracy came within 4.3 millihartree of density-matrix renormalization group, a leading classical simulation method, which the paper cites as evidence that quantum-classical hybrid workflows can approach the best classical methods on problems that remain extremely demanding for classical compute alone. For investors watching the quantum computing space, this provides a concrete, peer-reviewable data point rather than a roadmap promise, and it arrives as IBM separately had a collaboration with Cleveland Clinic and RIKEN named a finalist for the 2026 ACM Gordon Bell Prize.
Sources
Mentioned: NVDA