TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
NVIDIA says it simulated a 104-qubit fault-tolerant quantum processor on JUPITER, the European supercomputer at Forschungszentrum Jülich, using 8,192 GPUs. The simulation modeled magic-state cultivation under realistic hardware noise; NVIDIA says the approach it studied could require 90% fewer qubits than previous methods. The result is a computational study, not a demonstration of a working quantum processor.
NVIDIA announced that it simulated a 104-qubit fault-tolerant quantum processor on JUPITER, the supercomputer at Germany’s Jülich Supercomputing Centre, using 8,192 GPUs. The company says the simulation modeled magic-state cultivation under realistic hardware noise, offering researchers a way to study quantum processor designs before building them.
The work focused on magic-state cultivation, a method for producing the special quantum states needed to run fault-tolerant computations. According to NVIDIA, the approach studied could use 90% fewer qubits to produce magic states than previous methods. That reduction is a claim about the method, not a report that a physical quantum computer has already achieved the savings.
NVIDIA said the simulation used 8,192 GPUs built around GH200 superchips and ran with its cuQuantum software and the fault-tolerant logical layer of CUDA-Q. The company described the 104-qubit work as the largest comprehensive simulation of a fault-tolerant quantum processor to date. That characterization comes from NVIDIA; the announcement did not provide an independent comparison or third-party assessment.
The project was developed and tested before its run on JUPITER. NVIDIA said an GB200 NVL72 system at its Accelerated Quantum Research Center was used to scale preliminary tests from four GPUs to hundreds and compare results. The company said that development work helped reduce a simulation process that had taken weeks to days before the larger run.
Testing Quantum Designs Before Construction
Fault-tolerant quantum computers must correct or manage errors that arise because qubits are sensitive to noise. Producing magic states is a recurring part of many fault-tolerant approaches, so the resources needed to make them affect the scale and complexity of a future machine. Reducing that resource demand, if borne out in further research and hardware, could make designs more practical.
The immediate value of the JUPITER project is its use as a simulation and evaluation tool. Researchers can examine how a proposed design behaves under modeled noise before committing to a physical build. Such simulations do not remove the need for hardware experiments, but may help teams identify design issues and compare options earlier. NVIDIA and Jülich said this kind of work could support research aimed at applications including drug discovery, finance and materials science; the announcement does not report those applications being run on a quantum computer.
The result also illustrates the role of GPU supercomputers in quantum research: classical computing can model aspects of quantum processor behavior that are difficult to inspect directly. It does not mean JUPITER performed a quantum computation or that the simulated processor is ready for practical use.
quantum computing simulation software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Ideal Models to Noisy Simulations
The announcement contrasts the new work with earlier Jülich research that simulated 50 qubits under ideal, noiseless conditions. NVIDIA says the 104-qubit project instead modeled magic-state cultivation with realistic hardware noise, a more demanding test of how errors may affect a proposed design. The comparison concerns the simulation conditions and scale described by the organizations, not a direct test of two physical processors.
Large simulations can require substantial memory because their computational demands grow rapidly with system size. NVIDIA said its team prepared the methods on smaller GPU configurations and then scaled them to JUPITER’s 8,192 GPUs. JUPITER is identified in the announcement as Europe’s first exascale supercomputer, and the work was carried out at the Jülich Supercomputing Centre, part of Forschungszentrum Jülich.
“Useful quantum computing will require fantastically complex, fault-tolerant systems — and understanding how they behave at scale is essential to building them.”
— Jensen Huang, NVIDIA founder and CEO
GPU supercomputer for quantum research
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What the Simulation Does Not Establish
The announcement does not provide a peer-reviewed paper, independent verification of the 90% qubit reduction, or enough methodological detail to evaluate that comparison. It also does not specify the full noise model, simulation accuracy, or how the modeled performance compares with results from a physical quantum device. NVIDIA’s description of the run as the largest comprehensive simulation is a company claim, and the supplied material does not define the comparison set.
The simulation is not evidence that a 104-qubit fault-tolerant quantum computer was constructed or operated. It models a processor using classical supercomputing resources. How well the findings will translate to real hardware, and whether the proposed resource savings hold across different designs and error conditions, remain open questions.
fault-tolerant quantum processor kit
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Further Validation and Hardware Comparisons
NVIDIA and Jülich have not announced a publication date, independent validation plan or specific follow-up milestone in the material provided. Researchers can use the simulation methods to test further designs and compare their behavior under modeled noise, but the next meaningful test will be whether subsequent studies reproduce the results and whether hardware experiments support the design assumptions.
For now, the reported milestone is a large-scale classical simulation on JUPITER. Further details about the noise assumptions, accuracy and comparison with other approaches will be needed to judge how broadly the findings apply to future fault-tolerant quantum processors.
magic-state cultivation quantum hardware
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did NVIDIA simulate?
NVIDIA says it modeled a 104-qubit fault-tolerant quantum processor, focusing on magic-state cultivation under realistic hardware noise. The work ran on JUPITER using 8,192 GPUs.
Was a 104-qubit quantum computer built?
No. The announcement describes a classical computer simulation of a proposed quantum processor, not the construction or operation of a 104-qubit quantum device.
What does the 90% figure refer to?
NVIDIA says the magic-state cultivation approach studied could require 90% fewer qubits than previous methods. The supplied announcement does not include independent validation or full methodological details for that comparison.
Why use JUPITER for the work?
NVIDIA says large quantum simulations require significant computing resources as the modeled system grows. JUPITER’s 8,192 GPUs were used for the 104-qubit simulation after the team developed and tested its methods on smaller systems.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
