Nvidia unveils Ising AI models for quantum error correction and measurement

Technology and computing giant Nvidia Corp. today announced the release of Ising, the first open family of forensic models aimed at quantum computing simulation and error correction.

Nvidia, whose core business is graphics processing units that power AI, said these types of AI will allow researchers and business companies to build better quantum computers capable of running useful applications at scale.

To build and run useful applications, quantum computers must handle millions of qubits – atomic computing units of quantum information. A key problem is that qubits are fragile, error-prone and prone to noise at scale. As quantum computers develop, they must be error-corrected and calibrated in real-time to account for environmental factors and remain useful.

“AI is critical to making quantum computing work,” said co-founder and CEO Jensen Huang. “With Ising, AI becomes the control plane – the operating system of quantum machines – that transforms weak qubits into scalable and reliable quantum-GPU systems.”

Ising is named after historical mathematical model which helped simplify the understanding of complex physical systems by explaining how interacting particles, or spins, influence each other. Nvidia offers two models: real-time debugging and benchmarking.

The need for error correction is obvious: It turns noisy systems into coherent outputs. This is where the Ising emission comes into play. Decoding comes with two variants of a 3D convolutional neural network model, one optimized for speed and the other for accuracy, enabling real-time recording of quantum error correction. Nvidia said the models offer 2.5 times the speed and three times the accuracy pyMatchingthe current standard of the open source industry.

Ising Calibration allows physicists to configure systems by tuning, measuring and optimizing physical control signals, such as microwaves or lasers. This calibration is necessary to ensure high-fidelity results by correcting noise, hardware malfunctions and parameter degradation over time. It’s a visual language model that can quickly interpret and react to measurements from quantum processors, driving AI agents that change continuous measurements.

The path to better quantum computers

Speaking at the forum, Sam Stanwyck, Nvidia’s director of quantum products, said that the company chose to record and measure first because it faces the challenges that exist in the development of quantum systems.

He described both as “AI-shaped workloads,” where models can make an immediate impact today, but said Nvidia’s long-term vision is moving forward. Over time, the company expects AI to help design and optimize quantum circuits, making recording and measurement a milestone in the broader path to quantum-GPU-based supercomputers.

Ising Decoding and Ising Calibration are already accepted by business and research organizations. Decoding is distributed by Cornell University, Sandia National Laboratories, University of California San Diego and UC Santa Barbara, among others. The benchmark is already being used by Atom Computing, Academia Sinica, EeroQ, IonQ, IQM Quantum Computers, Q-CTRL and others.

Additionally, Nvidia has released a cookbook of instructions, including quantum computing workflows and training data, as well as the Nvidia NIM microservice. This will allow developers to customize, train, configure and model different hardware setups, and deploy them locally on researchers’ systems to protect sensitive data.

Image: SiliconANGLE/Microsoft Designer