NVIDIA has introduced Ising, a collection of open artificial intelligence models that the company describes as the world's first designed to accelerate the path toward useful quantum computers. Released on June 23, 2026, the models target two of the most stubborn problems standing between today's experimental quantum hardware and practical computation: quantum error correction and qubit calibration.
The launch signals NVIDIA's belief that classical AI, running on the GPUs it already dominates, will serve as the connective tissue that makes quantum machines viable. NVIDIA CEO Jensen Huang has framed the relationship bluntly, describing AI as the "control plane" of quantum machines. For more context on this story, see our ongoing more AI stories.
AI as the Quantum Control Plane
Quantum computers hold the promise of solving certain problems, such as molecular simulation and cryptography, far faster than classical machines. But today's quantum processors remain noisy and error-prone. Qubits, the basic units of quantum information, lose their delicate quantum states within fractions of a second, and keeping them stable requires constant, precise calibration.
That is where NVIDIA sees an opening. Rather than competing to build quantum processors itself, the company is positioning its AI software and GPU infrastructure as the layer that manages, stabilizes, and corrects the behavior of quantum hardware built by others. The Ising models are intended to help researchers and hardware makers apply machine learning to detect and correct errors and to tune qubits more efficiently.
By releasing the models as open weights, NVIDIA is encouraging the broader quantum research community to adopt, study, and improve them, a strategy that mirrors the open-weights approach that has reshaped the language model landscape.
Quantum Error Correction Meets Machine Learning
Quantum error correction is the field's central engineering challenge. Because quantum states collapse when observed directly, researchers must use indirect methods to detect and fix errors without destroying the information they are trying to protect. Conventional error-correction techniques demand enormous overhead, with many physical qubits needed to form a single reliable, or logical, qubit.
Machine learning offers a potential shortcut. AI models can learn to recognize patterns of errors and predict corrections from data, potentially reducing the overhead and latency of traditional decoding algorithms. NVIDIA's Ising models are built to operate in this gap, applying trained neural networks to the error-correction and calibration tasks that currently consume significant time and compute on quantum systems.
Early Real-World Results
The promise is already being tested in real hardware. Photonic quantum computing company Aegiq announced that it deployed NVIDIA's Ising AI models to automate calibration of its photonic quantum system, with reported results that AI cut the calibration time by a factor of three compared with previous methods.
For a field where hours of manual tuning are routine, that kind of reduction is significant. Faster, AI-driven calibration could let quantum hardware teams spend more time running useful computations and less time simply keeping their machines stable. Aegiq also combined the NVIDIA-accelerated approach with tensor network methods for large-scale fluid simulation, pointing to a hybrid model in which AI and quantum-inspired techniques work alongside one another.
A Broader Bet on AI for Science
The Ising launch is part of a wider NVIDIA push into scientific computing unveiled around the same period. At the ISC supercomputing conference, the company introduced new software libraries aimed at chemistry, materials discovery, and even the search for dark matter, alongside announcements that its technology now powers over 400 of the world's 500 fastest supercomputers.
NVIDIA also detailed new AI agent toolkits for life sciences and highlighted partnerships with national laboratories building supercomputers equipped with its latest CPUs. The common thread is a conviction that AI models, increasingly capable and increasingly open, can serve as general-purpose accelerators across the hardest problems in science, from drug discovery to quantum physics.
Quantum Stocks React
The market responded quickly to the announcement. Shares of quantum computing companies moved sharply higher following NVIDIA's debut of the Ising models, with quantum-focused stocks including D-Wave Quantum (QBTS), Rigetti Computing (RGTI), and Quantum Computing Inc. (QUBT) all posting notable gains. Investors interpreted NVIDIA's endorsement, and its open AI tooling, as a validation that the quantum sector is edging closer to commercial relevance.
The rally underscores how closely tied sentiment around quantum computing has become to the broader AI boom. When the leading GPU company releases models explicitly designed to make quantum hardware more practical, it reframes quantum not as a distant rival to classical computing but as an adjacent frontier that AI will help tame.
Why Open Weights Matter Here
NVIDIA's decision to release the Ising models as open weights is notable because quantum error correction is an area where progress has traditionally been siloed inside individual hardware companies and academic labs. Open models lower the barrier for researchers without access to proprietary decoders to experiment with AI-assisted correction.
If the open approach takes hold, it could accelerate the kind of collaborative, cross-institution progress that propelled large language models forward. Researchers can fine-tune the models on data from their own hardware, share improvements, and build a common toolkit rather than reinventing calibration and correction software for each new quantum platform.
The Road to Useful Quantum Computing
Quantum computing has spent years in a state of cautious optimism, with researchers repeatedly cautioning that fault-tolerant, broadly useful machines remain years away. NVIDIA is not claiming that Ising solves that problem outright. Instead, the company is betting that AI can compress the timeline by chipping away at the engineering bottlenecks, one calibration cycle and one corrected error at a time.
Whether that bet pays off will depend on how effectively the research community adopts the models and how much real-world performance they deliver beyond the early Aegiq results. But with Ising, NVIDIA has drawn a clear line: the future of useful quantum computing, in its view, will be built on a foundation of AI.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
