Alice & Bob has achieved a breakthrough in quantum error correction, slashing the logical error rate by a factor of 10,000 – without a proportional leap in qubit count. Published January 21, 2026, their new “Elevator Codes” utilize a novel approach to link error codes, employing a logical ancilla qubit that “moves” up and down other codes to detect bit-flip errors. This innovation builds upon the company’s existing cat qubit technology, partially protected from errors by design, and offers a pathway to dramatically more accurate quantum processors with a modest increase in hardware. “These error rates will make it possible to feasibly tackle problems like complex molecular simulation sooner than expected,” says Diego Ruiz, a theoretical physicist at Alice & Bob, hinting at the potential for accelerated progress in quantum computing.
Elevator Codes Reduce Bit-Flip Errors with Concatenation
Alice & Bob, a quantum hardware company based in Paris and Boston, is pioneering a novel approach to error correction dubbed “Elevator Codes,” detailed in the pre-print “Elevator Codes: Concatenation for resource-efficient quantum memory under biased noise.” The innovation centers on code concatenation – layering one error-correcting code atop another – to dramatically reduce bit-flip errors without a proportional increase in qubit count. This is particularly significant given the challenges of scaling quantum hardware due to the substantial overhead typically required for robust error correction. The core concept involves a logical ancilla qubit functioning like an “elevator,” dynamically moving between layers of repetition codes during computation. This movement facilitates checks for bit-flips at the logical level, enhancing the system’s ability to identify and correct errors. This efficiency stems from the codes’ high encoding rate, allowing more logical qubits to be packed into the same physical space, and the reuse of the ancilla qubit after each error check. Peter Shanahan, a quantum informatics researcher, and colleagues designed the codes to specifically address bit-flip errors, modifying the existing repetition code to extend its protective capabilities beyond phase-flips. Ruiz adds, “Also, this study is based on what our classical computers can simulate, it is reasonable to infer that even better performances could be reached using larger, higher rate codes on quantum hardware.”
Cat Qubit Architecture & Passive Error Protection
The innovation focuses on minimizing bit-flip errors—a significant hurdle in quantum computation—without a proportionate increase in qubit count. The company’s cat qubits possess inherent, or “passive,” protection against bit-flip errors by design. These Elevator Codes build upon this foundation by modifying the “active” protection typically used for phase-flips to also address bit-flips, creating a more robust system. Central to this advancement is code concatenation, layering a new code atop existing repetition codes. A logical ancilla qubit functions as an “elevator,” moving between these codes during computation to actively detect bit-flips at a logical level.
Adding a single logical ancilla qubit achieves a significant performance boost with minimal qubit overhead, enabled by resetting and reusing the ancilla after each check. Alice & Bob anticipates achieving a 10,000-fold reduction in the logical error rate while requiring only approximately three times more qubits compared to their current error correction methods.
Compared to Alice & Bob’s current error correction, the researchers expect to be able to reach a 10,000 times lower logical error rate while requiring only about 3 times more qubits using the new approach.
This ancilla “moves up and down” repetition codes, actively detecting bit-flip errors at the logical level—a crucial step in building reliable quantum processors. This efficiency is driven by a high encoding rate, allowing for denser packing of logical qubits. The design is particularly well-suited to Alice & Bob’s cat qubits, which already possess “passive” protection against bit-flip errors, and modifies existing repetition codes to broaden their protective capabilities.
Source: https://arxiv.org/abs/2601.10786
