Surprisingly Useful AI Article Enhancements
French-American quantum computing company, Alice & Bob, has demonstrated for the first time that its LDPC-cat error correction codes can be decoded on Nvidia GPUs, using the chipmaker’s NVQLink architecture to connect quantum control hardware with classical accelerators.
The proof of concept, which used synthetic data rather than a live quantum processor, represents a step toward fast, scalable error correction for cat-qubit quantum computers. Error correction remains one of the central bottlenecks in building fault-tolerant quantum machines, and the speed at which error signals can be decoded is a key constraint.
The Error Correction Problem
Quantum computers are inherently error-prone. Qubits, the basic units of quantum information, are susceptible to noise and interference that corrupt calculations. To build a quantum computer capable of running useful algorithms, those errors need to be detected and corrected in real time, a task that falls to classical computing hardware running alongside the quantum processor.
The challenge is that errors in current quantum systems accumulate faster than they can be corrected. The decoding step, where classical hardware interprets error signals from the quantum processor and determines what corrections are needed, must happen with minimal latency to keep pace with the quantum computation.
For faster qubit types such as cat qubits, the decoding demands are particularly acute.
What Alice & Bob Demonstrated
Alice & Bob assessed the feasibility of using Nvidia GPUs to decode its LDPC-cat codes via an NVQLink connection paired with Quantum Machines’ OPX1000 and Open Acceleration Stack. The Open Acceleration Stack is a modular framework that couples quantum control electronics directly to GPUs and other classical processors with microsecond-scale roundtrip latency.
The specific question the researchers set out to answer was whether measurement data could be transferred from the quantum control electronics to the GPU and decoded there using existing infrastructure.
The results, while based on synthetic data and not constituting real-time decoding, showed that decoding can become more efficient as multiple syndromes are decoded in parallel across the GPU. This parallelism is important because it suggests that decoding capacity can scale alongside the quantum processor as the number of qubits grows.
What Are LDPC-Cat Codes
LDPC-cat codes were developed by Alice & Bob specifically for its cat-qubit architecture and were first introduced in a paper titled “LDPC-cat codes for low-overhead quantum computing in 2D,” published in January 2025.
The codes are designed to reduce the number of physical qubits needed for fault-tolerant quantum computing by making error correction more efficient. In conventional approaches, the overhead in physical qubits required per logical qubit is substantial, often running into the thousands. LDPC-cat codes aim to compress that ratio by exploiting the noise-bias properties inherent to cat qubits.
Alice & Bob has previously claimed that its cat-qubit architecture could reduce the hardware requirements for building a useful large-scale quantum computer by up to 200 times compared with competing approaches.
The Role Of NVQLink
NVQLink is Nvidia’s open system architecture for coupling GPU supercomputing with quantum processors (QPUs). It is designed to support tasks that require fast communication between quantum and classical hardware, including error-correction decoding, calibration and hybrid quantum-classical applications.
The architecture is qubit-modality agnostic, meaning it is designed to work across different types of quantum hardware, whether superconducting, trapped ion, photonic or cat-qubit based.
In this proof of concept, NVQLink served as the communication layer between Quantum Machines’ control electronics and the Nvidia GPU performing the decoding.
What Comes Next
Alice & Bob’s next step is to optimise the full pipeline, the chain of steps that carries measurement data through NVQLink to the GPU, to achieve the performance required for real-time decoding during live quantum computation.
Related: Best Business Laptops for work & school
Related: Best Gaming Laptops
Related: Best Portable Laptop
US General Manager of Alice & Bob, Juliette Peyronnet, framed the work as an encouraging sign for the scalability of the company’s error correction approach.
“Error correction is what will turn quantum computers into useful machines, and decoding is one of its most demanding classical tasks,” Peyronnet explained. “Showing that our LDPC-cat codes can be decoded on Nvidia GPUs with existing tools is encouraging for scaling, because it means decoding capacity can grow with the machine. The next step is making it fast enough to keep pace with the quantum computer, and we’re working with Nvidia to get there.”
Director of Quantum Product at Nvidia, Sam Stanwyck, highlighted the significance of bringing GPU-accelerated decoding to faster qubit types.
“A major challenge with present day quantum computing is errors build up faster than they can be corrected, and traditionally it has been a challenge to bring state of the art compute to this problem for faster qubit modalities like cat qubits,” Stanwyck noted. “Alice & Bob’s work shows how NVQLink can in fact accelerate the real-time decoding needed to turn cat qubits into fault tolerant quantum-GPU supercomputers.”
An Ongoing Collaboration
The proof of concept builds on an existing relationship between the two companies. In March 2026, Alice & Bob and Nvidia reported a runtime of one hour and 57 minutes for decoding simulated syndrome data of its Elevator Codes using the Nvidia CUDA-Q platform. In June 2025, the companies announced the integration of Nvidia CUDA-Q into Dynamiqs, Alice & Bob’s QPU simulation library, enabling GPU-accelerated quantum simulations.
About Alice & Bob
Alice & Bob was founded in 2020 and is headquartered in Paris and Boston. The company has raised EUR 180 million in funding and employs more than 250 people. It specialises in cat qubits, a technology developed by the company’s founders, and its research is advised by Nobel Prize winning researchers.
Last Updated on October 13, 2026 by Nick Ross
