Researchers at the Jülich Supercomputing Centre and NVIDIA have fully simulated a universal quantum computer with 50 qubits, surpassing the previous 48-qubit record that Jülich scientists set in 2019 on Japan's K computer. The run took place on JUPITER, Europe's first exascale supercomputer, which officially launched at Forschungszentrum Jülich last September.

A 50-qubit simulation is not a modest step up from 48. Every added qubit doubles the memory and computing power the simulation needs, because a classical machine must track the full quantum state of the virtual processor. The result is a workload of roughly 2 petabytes of memory, about two million gigabytes, spread across thousands of nodes.

"Only the world's largest supercomputers currently offer that much," said Prof. Kristel Michielsen, Director at the Jülich Supercomputing Centre. "This use case illustrates how closely progress in high-performance computing and quantum research are intertwined today."

Why simulate a quantum computer at all?

Simulations let scientists test algorithms, check experimental results, and predict how future quantum systems will behave before the real hardware exists. Two algorithm families drive much of the interest:

• The Variational Quantum Eigensolver (VQE), used to study molecules and materials.
• The Quantum Approximate Optimisation Algorithm (QAOA), aimed at optimization problems in logistics, finance, and artificial intelligence.

The simulation does not stop at abstract math. It models the detailed quantum behavior of an actual processor, so every operation, such as applying a quantum gate, touches more than 2 quadrillion complex numerical values — a 2 followed by 15 zeros. Those values must stay synchronized across thousands of computing nodes to reproduce a real chip faithfully.

In plain terms: classical supercomputing has become the test bench on which quantum computing is being designed.

What made the record possible?

The hardware foundation is JUPITER's more than 16,000 NVIDIA GH200 Superchips. Each Superchip tightly couples a CPU and a GPU, which lets the simulation spill data that exceeds GPU memory into CPU memory without collapsing performance. A standard laptop, by comparison, can manage simulations of roughly 30 qubits.

The software side matters just as much. Engineers at the NVIDIA Application Lab, a joint effort between Jülich and NVIDIA, upgraded Jülich's quantum simulator JUQCS into a new version called JUQCS-50. It runs quantum calculations efficiently even when part of the data lives in CPU memory.

Two further techniques made the numbers work:

• A byte-encoding compression method that cuts memory requirements by a factor of eight.
• A dynamic optimization system that continually improves data exchange across the GH200 fleet.

"With JUQCS-50, we can emulate universal quantum computers with high fidelity and tackle questions that no existing quantum processor can yet solve," said Prof. Hans De Raedt of the Jülich Supercomputing Centre, lead author of the study, which was published as a preprint on arXiv on 7 November 2025.

Who gets to use it?

JUQCS-50 will not stay inside Jülich. Outside research organizations and companies will access it through JUNIQ, the Jülich UNified Infrastructure for Quantum Computing. The team expects it to serve both as a scientific tool and as a benchmark for evaluating future supercomputers.

The work ran under the JUPITER Research and Early Access Programme (JUREAP), which let hardware and software teams co-design during the machine's construction phase. "Through early collaboration, hardware and software could be co-designed during JUPITER's construction phase, in close cooperation between Jülich experts and NVIDIA — an important step towards realising the full potential of this exascale system," said Dr. Andreas Herten of the Jülich JUPITER project team, a co-author of the study.

How is JUPITER funded?

The machine draws on three public sources:

• Half from the European High Performance Computing Joint Undertaking (EuroHPC JU).
• One quarter from the Federal Ministry of Research, Technology and Space (BMFTR, formerly BMBF).
• One quarter from the Ministry of Culture and Science of the State of North Rhine-Westphalia (MKW NRW), through the Gauss Centre for Supercomputing.

The funding split shows that Europe treats exascale capacity as shared infrastructure rather than a national trophy project.

For now, the record confirms a practical point: the fastest path to debugging tomorrow's quantum machines runs through today's largest classical ones, and each new supercomputer generation will reset what quantum researchers can test.