Study Nvidia's record 104-qubit quantum sim on JUPITER for efficient error correction
Nvidia used the JUPITER supercomputer to simulate a 104-qubit quantum system that corrects its own errors, marking a new milestone for hardware testing.
Key points
- The simulation ran on JUPITER, Europe’s first exascale supercomputer located at Germany’s Jülich Supercomputing Centre.
- Nvidia’s method uses 90% fewer qubits for error correction than previous approaches by employing magic state cultivation.
- This represents the largest full simulation of a fault-tolerant quantum machine to date, according to the company.
Nvidia has demonstrated that conventional supercomputers can simulate complex quantum systems capable of correcting their own errors. The company used the JUPITER supercomputer in Germany to run this large-scale test. This work helps researchers evaluate quantum computer designs before building physical hardware. The simulation accounted for real-world noise that typically disrupts quantum calculations.
In plain English
Quantum computers use qubits, which are extremely sensitive to interference from their environment. Even minor noise from control equipment can cause calculation errors. A fault-tolerant system can detect and fix these errors while running. Most experts view this capability as essential for making quantum computers practically useful for solving complex problems.
The background
The simulation utilized Nvidia GH200 chips across 8,192 GPUs on the JUPITER system. JUPITER opened in September 2025 and is Europe’s first exascale supercomputer. It performs more than a billion billion calculations per second. This follows earlier work by Jülich researchers who simulated 50 qubits on the same machine in November 2025 under ideal, noise-free conditions.
What changes now
Nvidia developed the model on a smaller system at its quantum lab first. This initial testing cut the required work time from weeks to days. The entire simulation ran on conventional computers, not a quantum processor. Jensen Huang, Nvidia CEO, stated that useful quantum computing will require fantastically complex, fault-tolerant systems.
Key terms explained
Teams usually rely on simulations to validate hardware designs before committing to expensive physical builds. Check whether your organization uses similar modeling for critical infrastructure updates. Verify that test environments mirror real-world noise and interference levels. Ensure that error-correction protocols are validated in software before deployment. This approach reduces risk and allows for iterative refinement of complex systems without disrupting live operations or requiring immediate hardware access.
What to do and how to stay safe
- Review current hardware testing protocols to ensure they account for environmental noise and interference.
- Evaluate whether simulation tools are being used to validate complex system designs before physical implementation.
- Monitor vendor announcements for new error-correction techniques that could improve system reliability.
- Assess if your team has access to sufficient computing power for large-scale simulations of next-generation technology.
General security guidance from the Vector Update newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
How large was the quantum computer simulated by Nvidia?
Nvidia simulated a 104-qubit quantum computer,
