NVIDIA Ising: The Open Source AI for Quantum Processors

NVIDIA recently unveiled Ising, a family of open-source artificial intelligence models aimed at improving the reliability of quantum processors. These processors, while promising, suffer from chronic instability, making an error every thousand operations. NVIDIA aims to reduce this error rate by a factor of one billion using AI.

On April 14, Jensen Huang, CEO of NVIDIA, announced that Ising could become the "operating system" for quantum machines. This initiative marks a reversal of traditional roles, with AI now coming to the rescue of quantum computing.

Ising: A Two-Pronged Solution

Ising consists of two distinct modules. The first, Ising Calibration, is a vision-language model with 35 billion parameters. It is capable of analyzing the raw data from a quantum processor and autonomously adjusting its settings, a task that previously required several days of work from a team of engineers but can now be accomplished in just a few hours.

The second module, Ising Decoding, focuses on real-time error correction. NVIDIA claims that this module offers decoding that is 2.5 times faster and 3 times more accurate than current methods. These tools are available on GitHub and Hugging Face platforms, allowing laboratories to adapt them to their own architectures in just a few days.

Prestigious institutions such as Harvard, the Lawrence Berkeley National Laboratory, IQM Quantum Computers, and the UK’s National Physical Laboratory are among the first to adopt this technology.

Integration into the NVIDIA Ecosystem

Ising integrates into an ecosystem that NVIDIA has been developing for several years. The CUDA-Q platform, dedicated to hybrid quantum-classical computing, serves as the programming foundation, while NVQLink, introduced in October 2025, ensures the physical connection between GPUs and quantum processors. Ising adds a layer of intelligence to this framework, thus providing a complete technology stack.

The French ecosystem is already participating in this venture. Alice & Bob has been collaborating on NVQLink since 2025, and Pasqal has integrated its neutral atom processors into CUDA-Q. With Ising, these partners now have ready-to-use calibration and error correction tools, but this also implies an increased dependence on NVIDIA's infrastructure, a delicate issue for European technological sovereignty.

Open Source as a Strategy

NVIDIA's open-source initiative can be seen both as a gift and a strategy. By mastering the software layer of an emerging technology, NVIDIA secures a dominant position in the ecosystem, just as it did with CUDA in the field of AI. The quantum market, estimated to reach $11 billion by 2030, represents an opportunity that NVIDIA intends to seize, without necessarily producing quantum processors. While Google and IBM focus on developing qubits, NVIDIA prefers to provide the necessary tools for this technological revolution, thereby strengthening its strategic position.