NVIDIA Launches Ising AI Models for Quantum Computing

NVIDIA has unveiled a new family of open AI models aimed at tackling two of the biggest engineering hurdles in quantum computing: calibration and error correction. The models, called NVIDIA Ising, help researchers and enterprises develop quantum processors capable of running useful, large-scale applications.

For engineers and technology leaders following the quantum computing ecosystem, the announcement signals how AI is increasingly becoming a core layer in quantum system design. The move also highlights NVIDIA’s strategy to position GPUs and AI software as essential components in hybrid quantum-classical computing architectures.

AI moves into the quantum control stack

Named after the well-known Ising mathematical model used to describe complex physical systems, the new NVIDIA Ising family focuses on improving quantum processor reliability and scalability.

Quantum processors are notoriously fragile, requiring constant calibration and sophisticated error-correction techniques to keep qubits functioning correctly. According to NVIDIA, the Ising models provide AI-based tools to address both challenges, delivering quantum processor calibration capabilities and quantum error-correction decoding that can be up to 2.5× faster and three times more accurate than traditional approaches.

“AI is essential to making quantum computing practical,” said Jensen Huang, founder and CEO of NVIDIA. “With Ising, AI becomes the control plane — the operating system of quantum machines — transforming fragile qubits to scalable and reliable quantum-GPU systems.”

The company says the models help researchers tackle larger and more complex quantum workloads by improving the decoding process used in quantum error correction — a critical requirement for reliable quantum computing at scale.

Two core models: calibration and decoding

The NVIDIA Ising family includes two main components targeting different parts of the quantum stack.

• Ising Calibration is a vision-language model designed to interpret measurement signals from quantum processors. The model enables AI agents to automate the calibration process, reducing the time required from days to hours.
• Ising Decoding focuses on quantum error correction. It uses a 3D convolutional neural network architecture available in two variants optimized either for speed or accuracy. NVIDIA says these models outperform the current open-source standard, pyMatching, by delivering up to 2.5× faster performance and three times greater accuracy.

The models are designed to run locally on researchers’ systems, allowing organizations to maintain control of sensitive data while fine-tuning the models for their specific hardware platforms.

Early adoption across the quantum ecosystem

NVIDIA says several research labs, universities and quantum computing companies are already adopting the Ising models.

Organizations using Ising Calibration include Atom Computing, Academia Sinica, EeroQ, Conductor Quantum, Fermi National Accelerator Laboratory, Harvard John A. Paulson School of Engineering and Applied Sciences, Infleqtion, IonQ, IQM Quantum Computers, Lawrence Berkeley National Laboratory’s Advanced Quantum Testbed, Q-CTRL and the U.K. National Physical Laboratory.

Meanwhile, Ising Decoding is being deployed by groups including Cornell University, EdenCode, Infleqtion, IQM Quantum Computers, Sandia National Laboratories, SEEQC, the University of California San Diego, UC Santa Barbara, the University of Chicago, the University of Southern California and Yonsei University.

NVIDIA is also providing training data, workflow “cookbooks” and NIM microservices to help developers adapt the models for different quantum hardware architectures.

Part of a broader open-model strategy

The Ising release expands NVIDIA’s growing portfolio of open AI models, which already includes Nemotron for agentic AI systems, Cosmos for physical AI, Isaac GR00T for robotics and BioNeMo for biomedical research.

The models integrate with the company’s CUDA-Q software platform for hybrid quantum-classical computing and with the NVQLink QPU-GPU interconnect, aimed at enabling real-time quantum control and error correction.

With the quantum computing market projected to exceed $11 billion by 2030, according to analyst firm Resonance, progress in error correction and scalability will be key. NVIDIA’s approach suggests AI could become one of the main tools for overcoming those barriers.