Alice & Bob Introduce ‘Elevator Codes’ for Resource-Efficient Fault Tolerance

Alice & Bob Introduce ‘Elevator Codes’ for Resource-Efficient Fault Tolerance Alice & Bob has introduced “Elevator Codes,” a quantum error correction (QEC) architecture designed to minimize the physical hardware required for fault-tolerant computing. Detailed in a technical study authored by Diego Ruiz and Peter Shanahan, the architecture utilizes the inherent noise bias of cat qubits to address bit-flip and phase-flip errors at different layers. By concatenating classical repetition codes with high-rate outer codes, the design achieves a 10,000x reduction in logical error rates while requiring only 3x more physical qubits than the company’s current baseline repetition code. Standard fault-tolerant approaches often rely on 2D surface codes to correct both bit-flips and phase-flips. Alice & Bob’s cat qubits provide “passive” protection against bit-flips, reducing error correction to a 1D problem solvable by a repetition code. However, as computation scales, residual bit-flips eventually limit performance. Elevator Codes maintain 1D phase-flip protection as an “inner” layer but layer a high-rate “outer” code on top to manage bit-flips. This is achieved by “moving” a logical ancilla qubit—the “elevator”—vertically through the hardware to execute parity checks and detect errors without the massive overhead typically required by 2D surface code tiling. The architecture demonstrates substantial efficiency gains compared to established asymmetric codes such as the Rectangular Surface Code and the XZZX Code, particularly in high noise-bias regimes (η ≥ 7 × 10⁴). At a phase-flip error rate of 10⁻³, Elevator Codes reduce qubit overhead by over 50% to reach a logical error rate of 10⁻¹². For intermediate-term hardware (pZ = 10⁻²), the advantage is even more pronounced, offering a three-fold improvement in overhead compared to XZZX codes. This efficiency is critical for reaching “scientific-grade” accuracy (10⁻¹² to 10⁻¹⁵) required for complex molecular simulations. The development of Elevator Codes suggests that the timeline for running useful large-scale quantum algorithms may be shorter than previously estimated. By decoupling error types—prioritizing a high threshold for phase-flips in the inner layer and rate efficiency for bit-flips in the outer layer—the architecture circumvents the connectivity constraints of traditional 2D lattices. This allows Alice & Bob to target the computing power of 100 high-fidelity logical qubits with approximately 1,500 physical cat qubits, a footprint significantly smaller than that required by standard superconducting or ion-trap modalities. Read the official press release from Alice & Bob here and the technical preprint on arXiv here. January 21, 2026 Mohamed Abdel-Kareem2026-01-21T06:33:48-08:00 Leave A Comment Cancel replyComment Type in the text displayed above Δ This site uses Akismet to reduce spam. Learn how your comment data is processed.