In a significant development for the fields of quantum computing and cryptographic security, a multi-institutional team of researchers has released a comprehensive end-to-end resource estimation for running Shor’s algorithm using superconducting cat qubits. The study, detailed in a newly published paper on the Cryptology ePrint Archive (Paper 2026/2665), maps out the complete path from theoretical mathematical descriptions down to the concrete allocation and scheduling of physical qubits on a quantum processor chip layout.
Authored by Axel Pappalardo, Dahmun Goudarzi, Victor Hertz, Hugo Jacinto, Larissa Kroell, André Schrottenloher, Peter Shanahan, Élie Gouzien, Rémi de La Vieuville, and Vivien Londe—with researchers representing prominent organizations including quantum computing firm Alice & Bob, University Paris-Saclay, CEA, CNRS, University of Rennes, Inria, and the University of Edinburgh—the research tackles one of the most formidable practical challenges in quantum information science: realizing a cryptographically relevant quantum algorithm on physical hardware with realistic noise constraints.
A Comprehensive Framework for Modern Quantum Algorithms
The research provides a sweeping, unified view of modern adaptations of Shor’s algorithm, focusing specifically on three major cryptographic targets: RSA encryption, the Elliptic Curve Discrete Logarithm Problem (ECDLP), and the Finite Field Discrete Logarithm Problem (FFDLP). By examining these algorithms in depth, the authors make explicit the logical circuits that underpin them and carefully analyze the intricate hardware trade-offs they enable.
Rather than stopping at the abstract logical level, the research team successfully compiles the dominant subroutines down to physical-qubit operations across two distinct superconducting cat-qubit architectures: one predicated on repetition cat codes and another utilizing Low-Density Parity-Check (LDPC) cat codes. To achieve this, the authors engineered critical new hardware primitives. They designed a novel parity-CNOT gate tailored for cat qubits, extended a multi-target CNOT gate framework to function seamlessly within the LDPC-cat architecture, and adapted an active injection procedure specifically suited for cat-qubit systems.
Building directly upon these newly established primitives, the researchers adapted an advanced adder circuit originally introduced in Gidney’s 2025 resource estimate for factoring RSA-2048, modifying it to operate efficiently within cat-qubit hardware constraints. By combining this newly compiled adder with a coarser resource estimate for the phaseup operation, alongside factory designs and lookup table compilations derived from the FeMoco resource estimation framework previously published by Kroell and her colleagues, the team was able to generate fully realized chip layouts and operational schedules capable of factoring a 2048-bit RSA integer.
Resource Requirements and Performance Metrics for RSA-2048
The primary configuration fully compiled by the research team models a system operating under a one-microsecond error-correction cycle time, a ten-microsecond reaction time, and circuit-level phase-flip noise with a probability of $p_Z = 5 times 10^-4$, while assuming an ideal absence of bit-flips. Under these specific technical assumptions, factoring an RSA-2048 key requires approximately 23,000 cat qubits—specifically calculated at 22,593 qubits—and a total execution time of 3.14 days.
The study also explores the inherent flexibility of quantum algorithm compilation by demonstrating space-time trade-offs. For institutions or applications where processing speed is prioritized over physical hardware footprints, the researchers calculated that by allocating roughly 611,000 cat qubits, the RSA-2048 factoring process can be accelerated to run in just 1.25 hours.
Addressing the physical realities of quantum hardware noise, the researchers evaluated the strict baseline requirement of having no bit-flips, which mathematically corresponds to an underlying bit-flip time of approximately 100 years. To make the architecture more experimentally viable, they demonstrated that this stringent constraint can be relaxed significantly—down to a bit-flip time of about one hour—by incorporating an additional concatenation layer into a distance-3 bit-flip error-correcting code. When utilizing this mitigation strategy, the estimated hardware cost for factoring RSA-2048 increases moderately to about 45,000 cat qubits.
Expanding Benchmarks to Elliptic Curve and Finite Field Cryptography
Beyond the standard RSA benchmark, the research leverages the operations compiled for RSA-2048 to evaluate the resource requirements for solving other widely used cryptographic problems. Specifically, the authors examined the Elliptic Curve Discrete Logarithm Problem on the secp256k1 curve, a mathematical foundation extensively utilized in modern digital signature schemes and blockchain technologies. According to the study, solving ECDLP on secp256k1 requires at most 30,000 cat qubits and can be completed in approximately one hour.
Furthermore, the team proposed FFDLP-512 as a valuable intermediate benchmarking standard for the quantum computing community. Under their compiled architecture, solving FFDLP-512 requires approximately 6,700 cat qubits with an execution time of roughly 21 minutes. These expanded estimates illustrate the versatility of the compilation techniques developed by the authors, offering a clear roadmap for benchmarking quantum processor performance across diverse cryptographic domains as hardware development continues to advance.

