Logical qubits have achieved 96% survival rate: a breakthrough in quantum error correction on IBM Heron
The quantum industry is taking another important step toward fault-tolerant computing. A group of researchers from the University of Sydney, in collaboration with IBM engineers, has achieved a significant increase in the preservation of logical qubits — up to 96% per error correction cycle. The experiment was conducted on the latest 156-qubit superconducting processor, IBM Quantum Heron r2, and the results are impressive.
The main enemy of quantum systems on the path to the FTQC (fault-tolerant quantum computing) era remains the so-called "idle noise." The problem is that for error correction, the processor must regularly pause computations for internal checks. It is during these pauses that the remaining qubits lose stability and generate new errors, negating the correction efforts.
To solve this fundamental problem, physicists completely redesigned the architecture of the correction circuits. Instead of tolerating long idle periods, they radically reduced the computation pause time by optimizing the algorithms. The result is clear: the survival rate of logical qubits jumped from less than 90% to 96%.
Project leader Stephen Bartlett from Sydney Nano rightly notes that "forced idle time" is not just a technical detail, but a serious obstacle to reliable operation at every stage of computation. And although the current result was obtained in laboratory conditions on a single processor, its significance for the industry is hard to overestimate.
It is scalability and fault tolerance that remain the last bastions on the path to practical quantum supremacy. Recall that IBM previously announced plans to achieve the first confirmed cases of quantum advantage by the end of 2026. This achievement is another brick in the foundation of that ambitious goal.
Analyst's opinion: The increase in survival rate from 90% to 96% is not just numbers. In quantum error correction, every percentage point has exponential significance. If this method can be scaled to thousands of qubits, we will witness the transition from experimental machines to truly useful computing systems.