Industry Adopts Ising

NVIDIA’s new Ising artificial intelligence models are already moving from launch slide to lab workflow: Infleqtion said it will use them to tune its quantum hardware. (nvidianews.nvidia.com) NVIDIA announced Ising on April 14, 2026 as an open model family for two jobs that slow quantum computing down: calibrating processors and correcting errors while they run. NVIDIA said the package includes base models, training tools, data sets, and deployment workflows. (nvidianews.nvidia.com) (developer.nvidia.com) Infleqtion was on NVIDIA’s launch list of adopters alongside Harvard University, Fermi National Accelerator Laboratory, Lawrence Berkeley National Laboratory’s Advanced Quantum Testbed, IQM Quantum Computers, Academia Sinica, and the United Kingdom’s National Physical Laboratory. (nvidianews.nvidia.com) Quantum processors use qubits, which are information units that are powerful but fragile, and NVIDIA said top systems still make an error about once every thousand operations. The company said useful machines will need error rates closer to one in a trillion, which makes calibration and error correction central engineering problems. (developer.nvidia.com) Calibration is the process of measuring how a quantum chip drifts and retuning it, like repeatedly tuning an instrument that slips out of pitch. NVIDIA said its Ising Calibration model is a 35 billion parameter vision-language model trained on qubit data to automate that work. (developer.nvidia.com) Error correction works like a spotter watching for mistakes faster than they pile up, except the spotter is a classical computer attached to a quantum machine. NVIDIA said its Ising decoder models can run quantum error-correction decoding up to 2.5 times faster and with up to 3 times higher accuracy than traditional approaches. (nvidianews.nvidia.com) (developer.nvidia.com) Infleqtion had already been building toward that kind of hybrid setup with NVIDIA. In October 2025, the company said its neutral-atom Sqale quantum computer would connect to graphics processing unit systems through NVIDIA NVQLink for microsecond-scale feedback loops needed for calibration, decoding, and control. (infleqtion.com) Infleqtion repeated that strategy at NVIDIA GTC 2026 in March, saying Sqale would be shown as part of an NVQLink demonstration for real-time quantum error correction and hybrid artificial intelligence workloads. The company said its hardware would work with NVIDIA graphics processing units inside an accelerated high-performance computing environment. (ir.infleqtion.com) NVIDIA is also pitching Ising as open infrastructure rather than a closed service. The company said users can fine-tune models for their own hardware while keeping proprietary quantum processor data on-site, a point that matters for companies building custom devices with different noise patterns. (developer.nvidia.com) The immediate test is whether early adopters can turn those models into steadier qubits and faster feedback on real machines. Infleqtion’s decision gives NVIDIA an industry user to point to as that test begins. (nvidianews.nvidia.com)