Key Notes
- NVIDIA, IBM and Fermilab plan a joint platform to study interconnected quantum computers and demanding physics problems.
- NVIDIA reports up to 13,000x faster decoding for quantum error-correction research, a specific classical computing workload.
- Separately, 192-processor Vera CPU racks are expected from hardware partners beginning in early 2027.
NVIDIA, IBM and Fermilab are planning a research platform that would pair quantum processors with accelerated computing to help design larger, interconnected quantum machines. The collaboration targets demanding physics calculations and the engineering needed to make future quantum computers reliable.
NVIDIA announced the initiative on November 16 alongside a new error-correction decoder. Separately, the company unveiled rack-scale Vera CPU systems for scientific computing, expanding its push into the infrastructure used to model climate, molecules and other complex systems.
Building Quantum Computers Around Scientific Problems
The proposed platform would co-locate NVIDIA DGX B200 systems with IBM quantum computers, including systems powered by its 120-qubit Nighthawk processor. IBM outlined plans to use that combination to study how quantum processors and conventional computing resources can scale together.
Fermilab would bring researchers working on high-energy and nuclear physics into the design process. Potential applications include neutrino interactions, particle physics and nuclear simulations. The aim is to turn the requirements of those calculations into practical decisions about hardware, algorithms and software.
A central question is how separate quantum modules can cooperate. Fermilab said researchers would initially model and emulate the links between modules, measuring the effects of communication errors, delays and feedback. Those results would guide work on physical interconnects.
In a later phase, Fermilab intends to add superconducting cavity hardware from its Superconducting Quantum Materials and Systems Center. Researchers would explore uses including quantum memory and error correction, alongside AI-assisted optimization and real-time control.
The announcement describes a development program. It does not establish that the partners have already built a scalable, fault-tolerant quantum computer, and the releases give no launch date for the proposed platform.
What the 13,000x Decoding Claim Measures
NVIDIA also introduced TCBP-JOBS, a decoder available in CUDA-Q QEC 0.9.0. Decoders interpret measurement data to infer likely errors, an essential part of protecting quantum information.
The company reports speedups of up to 13,000x over leading existing algorithms in computationally intensive error-correction research. That is a claim about a particular classical computing task used to develop quantum systems. It is not a 13,000x acceleration of every quantum calculation or evidence that fault tolerance has been solved.
Faster decoding research could let teams evaluate more error-correction designs. It addresses a different part of the reliability challenge from Google’s Willow work, which demonstrated improvements in preparing low-error quantum states.
IBM and NVIDIA are also working to make quantum processors easier to use within existing supercomputing workflows. An integration involving IBM’s Quantum Resource Management Interface and the Slurm scheduler would let computing centers allocate and account for quantum processor time alongside CPU and GPU resources.
Vera CPU Racks Target Scientific Computing
In its separate hardware announcement, NVIDIA introduced a liquid-cooled rack combining 192 Vera CPUs, approximately 300TB of memory and Quantum-X800 InfiniBand networking. Each CPU has 88 custom Olympus cores, with the platform aimed at data-intensive simulations and AI workflows.
NVIDIA says its OpenFOAM airflow simulation tests delivered 1.9x the performance and 2.4x the performance per watt of the x86 systems it compared against. These are company benchmark results for a specific workload.
Dell Technologies, HPE, Lenovo and Supermicro are expected to offer Vera CPU racks beginning in early 2027. The German Climate Computing Center has also selected Vera CPUs for its HLRE-5 supercomputer, expected in early 2028. These are separate deployments from the proposed quantum platform, whose announced conventional computing component is DGX B200.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.