Quantum computers need a way to make fragile information reliable. On December 9, 2024, Google introduced Willow, a 105-qubit processor, alongside results showing a key step in quantum error correction: increasing the size of an encoded qubit reduced its error rate. The research paper describes operation below the surface-code error threshold.
Making more qubits work together
A logical qubit spreads information across multiple physical qubits, with repeated checks used to detect errors. Adding hardware helps only when the underlying errors are sufficiently low. In Google’s explanation of the experiment, increasing the code size from 3×3 to 5×5 to 7×7 reduced the encoded error rate at each step. This supports a route toward more reliable computation, while leaving substantial scaling work.
What the five-minute benchmark means
A separate random-circuit-sampling experiment produced the widely shared comparison: under five minutes on Willow versus an estimated 10 septillion years on a leading classical supercomputer. That figure was Google’s estimate for a particular benchmark, dependent on classical simulation assumptions. It was not a measurement of speed across ordinary applications or a demonstration of a useful drug-discovery calculation.
Why the distinction matters
The benchmark tests a demanding quantum task; error correction addresses whether useful computations can remain reliable as systems grow. In its launch report, Google explicitly described useful, beyond-classical applications as a next challenge. Willow’s 2024 results support progress toward that goal. They do not establish a general replacement for classical computers or the end of experimental quantum engineering.
Sources reviewed September 24, 2026. This article explains the December 2024 results, not the current performance of every quantum platform.
