Announcement
NVIDIA announced that it has completed a 104‑qubit simulation of a fault‑tolerant quantum processor on JUPITER, Europe’s first exascale supercomputer hosted at the Jülich Supercomputing Centre in Germany. The simulation is described as the largest comprehensive fault‑tolerant quantum‑computing system ever run.
Technical Details
The work employed NVIDIA’s cuQuantum software and the CUDA‑Q platform’s fault‑tolerant logical layer. It demonstrated magic‑state cultivation, a fault‑tolerant approach that reduces the number of physical qubits required by roughly 90% compared with traditional distillation protocols, which typically consume 90% to 95% of a quantum computer’s qubit budget to produce T‑states. Unlike earlier Jülich work that simulated only 50 qubits under ideal, noiseless conditions, this run modeled magic‑state cultivation under realistic hardware noise, providing a quantitatively grounded tool for predicting future processor behavior before hardware is built.
Development began on an NVIDIA GB200 NVL72 system at the NVIDIA Accelerated Quantum Research Center (NVAQC), where tests were scaled from 4 GPUs to several hundred before being transferred to JUPITER’s full 8,192‑GPU configuration. This staged approach compressed simulation time from weeks to days, according to NVIDIA’s blog.
Competitive Landscape
On the same day, Google Quantum AI published results in Science showing the first physical hardware implementation of magic‑state cultivation on its Willow superconducting processor, achieving a 40‑fold error reduction and a T‑state logical fidelity of 99.99(1)%. Quantinuum (NASDAQ: QNT) announced a world‑record quantum error‑correction result on its Helios commercial system, attaining a logical fidelity of 99.9995%, a ten‑fold improvement, as it targets a universal fault‑tolerant machine named Apollo by 2029. These parallel milestones indicate rapid progress across multiple quantum architectures.
Market Reaction
NVIDIA shares were trading at $217.42, up 1.54% in the Monday session following the announcement.
Executive Comments
Jensen Huang, founder and CEO of NVIDIA, said in a blog post that “useful quantum computing will require fantastically complex, fault‑tolerant systems, and understanding how they behave at scale is essential to building them,” adding that the results show how quantum processors and GPU supercomputers can collaborate on research otherwise out of reach. Kristel Michielsen, director of the Jülich Supercomputing Centre at Forschungszentrum Jülich, noted that the partnership brings a new capability to quantum computing, helping scientists make better‑informed design decisions.
Outlook
NVIDIA did not disclose a timeline for making the fault‑tolerant simulation tools publicly available to external quantum hardware partners, leaving the potential for cuQuantum and CUDA‑Q to become standard development environments an open question.