JÜLICH, October 26, 2026 — Nvidia and researchers at Germany’s Jülich Supercomputing Centre have fully simulated a 50-qubit universal quantum computer on classical hardware, marking the first achievement of its kind for Europe. This record run took place on JUPITER, the continent’s first exascale supercomputer, which was inaugurated in September 2025.
The Shift in Quantum Benchmarking
Europe’s flagship HPC center is now using massive classical power to benchmark quantum utility without physical noise. Classical simulations behave like perfect quantum computers with zero interference, providing a noise-free sandbox for testing algorithms and error correction methods. This approach allows researchers to validate designs before building expensive hardware that operates at the limits of physics.
Every qubit added to a simulation doubles the amount of information the classical machine has to track. Small gains in qubit count therefore demand outsized jumps in memory and compute resources. The previous high-water mark was a 48-qubit simulation set in 2019 using Japan’s K computer.
This new milestone handles a state vector of more than 2 quadrillion complex amplitudes. For Nvidia, the result reinforces its positioning as central to quantum computing workflows without building a quantum computer itself. The GH200’s hybrid memory design was a key ingredient, offering a concrete showcase for how memory architecture determines how far a quantum simulation can scale.
JUPITER Hardware and Memory Architecture
The run relied on JUPITER being built on Nvidia GH200 Grace Hopper Superchips. These chips pair a CPU and a GPU with hybrid memory access, allowing processors to share a larger pool of memory than a GPU could reach on its own. The nodes are tied together with Nvidia’s Quantum-X800 InfiniBand networking for high-speed data transfer.
For Europe, the run is an early proof point for JUPITER as part of the exascale computing era. It shows how classical supercomputers can accelerate quantum research by simulating error correction and magic-state distillation under realistic noise conditions.
What This Means Now
This work demonstrates that useful quantum computing will require fantastically complex, fault-tolerant systems. By running these simulations on JUPITER, researchers are getting the data needed to predict performance before building hardware, cutting reliance on trial and error. The shift is clear: classical supercomputers are the bridge to practical quantum applications in fields like drug discovery, financial modeling, and materials science.
Watch this space as more institutions use JUPITER to push the boundaries of what a fault-tolerant quantum processor can do. If you’re looking for the latest in quantum computing hardware or simulation tools, keep an eye on how these classical benchmarks evolve into real-world quantum advantage.
Jülich Supercomputing Centre and European HPC Leadership
Forschungszentrum Jülich, commonly referred to as Jülich, is a major research institution located in Germany and one of the largest in Europe, operating under the Helmholtz Association with roughly 6,800 employees across ten institutes and 80 subinstitutes. The institution traces its origins to 11 December 1956, when the State Parliament of North Rhine-Westphalia decided to establish an “atomic research centre,” and it adopted its current name in 1990. Jülich operates a broad range of research infrastructures, including supercomputers, and has long been a focal point for interdisciplinary research in energy, information, and bioeconomy.
Jülich Supercomputing Centre, a department within Forschungszentrum Jülich, plays a central role in European high-performance computing. The centre houses JUPITER, Europe’s first exascale supercomputer, which was inaugurated in September 2025. JUPITER is built on Nvidia GH200 Grace Hopper Superchips interconnected via Nvidia Quantum-X800 InfiniBand networking. The JUPITER project is further supported by the JUPITER Research and Early Access Program (JUREAP) and is integrated within JSC’s JUNIQ quantum user facility, reinforcing Jülich’s position as a leading hub for cutting-edge computational research in Europe.
JUPITER’s Exascale Hardware and Nvidia GH200 Superchips
The record-breaking 50-qubit simulation on JUPITER relied on Nvidia GH200 Grace Hopper Superchips, which pair a CPU and a GPU with a unified pool of high-bandwidth memory. This hybrid memory architecture was essential, as every additional qubit doubles the classical state vector size, requiring exponentially more memory. For the 50-qubit run, JUPITER handled a state vector of more than 2 quadrillion complex amplitudes. The nodes are interconnected using Nvidia Quantum-X800 InfiniBand networking, enabling high-speed data transfer necessary for distributing the massive computational workload across thousands of processors. This hardware configuration allowed researchers to overcome the memory bottlenecks that previously limited simulations to 48 qubits, demonstrating how exascale supercomputers can serve as powerful tools for quantum algorithm development.
According to the project results, the JUQCS-50 simulator achieved a speedup of approximately 16.6 times compared to the previous 48-qubit benchmark, thanks to three key software upgrades running on the GH200 hardware. Enhanced CPU offloading allowed the system to push workloads onto CPU memory when GPU memory was exhausted, while an eightfold compression of memory requirements and smarter network traffic management between nodes further improved efficiency. The GH200’s design was highlighted as a key factor in this scaling, offering a concrete example of how memory architecture determines the scalability limits of quantum simulations. This performance gain reinforces Nvidia’s role in providing the computational infrastructure needed to validate quantum designs before physical hardware is built.
JUQCS-50 Software Optimizations and Scaling
The Jülich Universal Quantum Computer Simulator, upgraded to version JUQCS-50, incorporated three key software upgrades that delivered a speedup of approximately 16.6 times compared to the previous 48-qubit benchmark. Enhanced CPU offloading allowed the system to push workloads onto CPU memory when GPU memory was exhausted, while an eightfold compression of memory requirements and smarter network traffic management between nodes further improved efficiency. According to the project results, the GH200’s hybrid memory design was a key factor in this scaling, offering a concrete example of how memory architecture determines the scalability limits of quantum simulations. This performance gain reinforces Nvidia’s role in providing the computational infrastructure needed to validate quantum designs before physical hardware is built.
The simulation utilized 8,192 GPUs and NVIDIA’s quantum computing software to help researchers assess processor designs before building hardware. Running on a system of NVIDIA GH200 superchips at Jülich Supercomputing Centre, the JUQCS-50 simulator modeled how a realistic quantum processor performs fault-tolerant computation at record scale. The approach studied could reduce the qubits needed to produce magic states by 90% compared with previous methods, according to NVIDIA. Because qubits operate at the limits of physics, the interactions needed to control them introduce unavoidable, error-inducing noise; computing in the presence of such noise—known as fault-tolerant quantum computing—requires the use of magic states, a specific arrangement of qubits that is continually consumed during a fault-tolerant quantum computation. Reducing the resources needed for useful quantum computing, the simulation validated a state-of-the-art magic-state cultivation approach, giving researchers the data and tools needed to predict the performance of future designs before they are built.
Sources
- Nvidia and Jülich simulate a 50-qubit quantum computer on classical hardware
- Forschungszentrum Jülich
- NVIDIA Runs Simulation of Fault-Tolerant Quantum Computing at Jülich Supercomputing Centre
Photo: Pimvantend (CC BY-SA 4.0)
