However, system characterization exhibited some major limitations. Cat Qubits is a bit flip, but not impossible. By focusing error correction only on phase flips, the bit flips that occur avoid inevitably causing logical, error-corrected qubit overall failures. “It is difficult to achieve long logical bit-flip times, as a single CAT kit bit-flip event in any part of the repeating code causes a direct logical bit-flip error,” the author points out. Another problem is that the transmon used for error correction is still affected by both bit flips and phase flips.
Where does this leave us?
Many companies, like Amazon, bet that by using error-prone hardware Qubit, you can use less total hardware Qubits to get effective error correction. If they are correct, they can build error-corrected quantum computers with much fewer qubits, which could result in faster useful calculations. For them, this paper is an important validation of an idea. A kind of mixed-mode error correction can be done using robust hardware Qubit, paired with compact error correction codes.
But beyond that, the message is quite mixed. Hardware had to rely on robust hardware qubits (transmons) to perform error correction, and very low error rates were not so low that they avoided occasional bit flips. And ultimately, the improvement in error rates obtained by increasing the size of the logical Qubit is not on the trajectory of obtaining useful levels of error correction without the need for a large number of hardware kibits.
In short, the underlying hardware is currently not sufficient to allow any kind of complex calculation, and radical improvements are required before it becomes possible. And there is no obvious alternative route to effective error correction. There is still a possibility for this approach, but it is not clear how to build the hardware to reach that potential.
As for Amazon, the photos aren’t even clearer given that this is the second Qubit technology we’ve spoken publicly. It is unclear whether the company is looking for a technology they want to commit to.
Nature, 2025. doi: 10.1038/s41586-025-08642-7 (About DOI).