Alice and Bob Quantum
The race to fault-tolerant quantum computing has advanced significantly with the announcement of a thorough integration with NVIDIA CUDA-Q technology by the Paris-based startup Alice & Bob. This partnership has shown a startling 75x acceleration in the simulation of quantum error correction (QEC) protocols, based on Alice & Bob’s exclusive “Dynamiqs” simulation library. For researchers trying to get over the “noise” barrier that now besets quantum electronics, this discovery offers a crucial tool. A new era of “intelligent” quantum architecture is made possible by the 75x performance improvement, which is more than simply a statistic.
You can also read How Alice&Bob Company Solves Quantum Noise with Cat Qubits
The Decisive Battle Against Decoherence
One must first comprehend decoherence, the fundamental obstacle of the quantum age, in appreciate the gravity of this news. Quantum bits, or qubits, are infamously brittle in contrast to classical bits, which are stable. Calculation mistakes can result from a qubit losing its state due to even the smallest environmental disturbance, such as heat, vibration, or electromagnetic waves. Quantum computers must attain an error rate of about one in a billion in address practical issues in domains such as chemistry, materials research, or cryptography. On the other hand, current physical qubits fail roughly once per 100 operations.
Quantum Error Correction (QEC), which combines numerous “noisy” physical qubits into a single “clean” logical qubit, is the remedy for this instability. Errors can be found and fixed using this method without measuring the quantum information directly and destroying it. However, because the complexity of the simulation increases exponentially with the number of qubits, replicating these error-correction cycles on classical computers is a computational nightmare. The partnership between NVIDIA and Alice & Bob becomes groundbreaking at this moment.
You can also read Unfolded Distillation: Alice & Bob and Inria’s Quantum Leap
The “Cat Qubit” Innovation
What makes Alice & Bob most renowned is their groundbreaking work on “cat qubits,” a specific type of superconducting qubit named after Schrödinger’s cat. Because of their “passive” protective design, these qubits are naturally immune to bit-flip mistakes. They still need “active” correction for phase-flip faults, even if this offers them a significant advantage in the quest for stability.
The business created Dynamiqs, an open-source high-performance simulation toolkit, to improve these active correction techniques. Dynamiqs may now delegate the demanding mathematical work of quantum dynamics to NVIDIA’s potent GPUs by integrating CUDA-Q, an open-source framework for hybrid quantum-classical computing. This integration enables simulations up to 75 times faster than current industry-standard libraries, according to early benchmarks. This implies that a researcher may now finish a simulation that used to take a week in just one afternoon.
You can also read Alice & Bob, Qilimanjaro New Tech at APS 2026 March Meeting
From Gaming to Scientific Computing
The initial aim of the hardware architecture that is currently driving this quantum revolution was rather different. GPUs were developed in the 1990s to use huge parallelism to generate hyper-realistic worlds in video games. GPUs carry out thousands of smaller computations at once, as opposed to traditional CPUs’ sequential computation. NVIDIA CUDA was introduced in 2006 as a result of academics’ realization in the early 2000s that these processors could be “tricked” into performing scientific computing.
With the advent of NVIDIA CUDA-Q in the 2020s, quantum computers can now collaborate with traditional computation resources like GPUs and CPUs. In actuality, operating a quantum computer will probably require a “double act” in which both quantum and classical hardware are used to address the particular aspects of issues where each is superior. According to Renéau Peronnin, CEO of Alice & Bob, improved simulations can instantly speed up the development of their Quantum Processing Units (QPUs), and simulation is a crucial step in the creation of functional processors.
You can also read HPC SLURM News: Now Support Cat Qubit Quantum Workloads
Case Study: The Elevator Codes
Decoding is a crucial step in evaluating the performance of quantum codes. Since errors cannot be directly observed in a true quantum device, researchers monitor observables known as stabilizers, which result in classical bits known as a syndrome. This syndrome suggests that something went wrong, although it’s unclear exactly where. In simulations, researchers employ classical software to identify the most likely fix after creating mistake patterns based on a noise model.
Recently, Alice and Bob benchmarked their “Elevator Codes” a novel error correction algorithm created to take use of cat qubits’ highly biased noise. The goal of these programs is to achieve the incredibly low logical error rates required for complex simulations such as FeMoco, a chemical essential to nitrogen fixation. The team used NVIDIA’s Grace Hopper to simulate 100,000 “shots” of the Elevator algorithm 9.25 times faster than a high-end CPU-only system. The CPU took 18 hours and 2 minutes to finish the work, whereas the GPU-based method finished it in 1 hour and 57 minutes.
You can also read The CSIS Center for Strategic and International Studies News
The 200x Advantage and the Path Ahead
Beyond only efficiency, this speed has ramifications for Quantum Optimal Control, which allows researchers to precisely “guide” systems toward a goal state by computing gradients of simulation outputs. Additionally, Alice and Bob are getting closer to the “holy grail” of real-time decoding where the classical controller processes error data quicker than the quantum hardware generates it with NVIDIA’s low-latency link.
According to Alice & Bob’s roadmap, they might reduce the hardware requirements for a working quantum computer by 200 times when compared to competitors’ traditional methods. While traditional approaches could require 20 million physical qubits to run Shor’s algorithm, Alice & Bob aims to achieve the same result with roughly 100,000 cat qubits. The company is successfully reducing the time to “Quantum Supremacy” for real-world applications by simulating and improving these architectures using NVIDIA’s GPUs.
With the utilization of NVIDIA DGX Quantum, which integrates Grace Hopper Superchips with control systems to form a tight loop between the GPU, CPU, and QPU, integration into the datacenters of the future is already under way. The combination of NVIDIA’s accelerated platform and cat qubit architecture gives Alice & Bob a major competitive advantage as the race for the first universal, fault-tolerant quantum computer intensifies. The objective is still very clear: to create a computer that can correct qubits more quickly than ever before in solve the most difficult issues in the world.
You can also read At NQCC, Infleqtion unveils UK’s 100 qubit quantum computer
