Google’s Willow processor completed a demanding quantum speed test in under five minutes. Google estimated that Frontier would need about 10 septillion years to reproduce the same random circuit sampling task under the classical simulation assumptions used in its comparison.
That contrast makes Willow’s computational reach tangible: the chip executed defined quantum circuits that the best classical methods in Google’s model could not reproduce on a practical timescale. Willow also advanced the engineering needed for longer calculations, because a larger surface code reduced logical errors instead of accumulating them faster.
Five minutes on Willow, 10 septillion years in Google’s estimate
Random circuit sampling gives a processor a demanding sequence of quantum gates and asks it to produce samples from the resulting distribution. The task stresses the whole system: gate quality, connectivity, calibration, control, and the ability to preserve quantum correlations all affect the result.
This test gives engineers a concrete answer to one question: can the chip execute circuits whose output distribution has become impractical to reproduce with the best known classical methods? Google reported that Willow finished its RCS run in less than five minutes and estimated about 10 septillion years for Frontier to perform the comparison task.
The estimate depends on the chosen circuit, target fidelity, classical algorithm, memory, storage, and hardware assumptions. Google included several memory scenarios and acknowledged that classical simulation methods will continue to improve. The result establishes Willow’s computational reach under a defined test. The next milestone is to direct comparable reach toward a scientific calculation that produces useful new information.
Larger codes protected quantum information more effectively
A physical qubit is fragile, so error correction spreads one logical unit of quantum information across many physical qubits. The code succeeds when adding more physical qubits lowers the logical error rate. Researchers call this operating below threshold.
On Willow, increasing the code distance by two reduced the logical error rate by a factor of 2.14. In the reported measurements, the distance-7 memory used 101 qubits, reached an error rate of 0.143% per correction cycle, and preserved quantum information 2.4 times longer than its best individual physical qubit, while a separate distance-5 system averaged 63 microseconds of decoder latency for correction cycles running every 1.1 microseconds.
These measurements give engineers a scalable direction: better physical components and larger codes can jointly produce a more dependable logical memory. Other teams are also exploring different routes toward fault-tolerant quantum computers, because useful quantum algorithms will require logical operations that remain accurate for far longer than today’s experiments.
The 47-year result marks an earlier step with Sycamore
The estimate of 47 years belongs to an earlier Google processor and helps show how this benchmark evolved. In August 2023, Google described a Sycamore RCS experiment with 700 two-qubit gates and a measured fidelity of 1.7 x 10^-3. The company estimated about 47 years of classical simulation on the world’s largest supercomputer.
The peer-reviewed version later reported a 67-qubit experiment running 32 circuit cycles. The researchers studied how noise changes the structure of the output and identified a weak-noise phase in which the circuit uses the processor’s full computational space and resists known classical shortcuts.
The SciencePost story published in January 2026 returned to this Sycamore result. In the timeline, the 47-year estimate describes the 2023 experiment, while Willow’s much larger estimate arrived later with a different chip. Both numbers are tied to specified RCS tests and to the classical methods available for estimating their simulation cost.
Error correction opens the path to longer useful calculations
Willow’s central contribution is a tested way to make encoded quantum memory more reliable as the code grows. The Nature paper also identifies the next engineering work: scale the hardware, control rare correlated errors, decode larger streams of data, and perform logical operations rather than memory tests alone.
The authors reported rare bursts in their repetition-code experiments and explained that large fault-tolerant algorithms still require further scaling. An April 2026 author correction fixed labels in one figure without changing the reported conclusions.
Quantum computation is also distinct from sending quantum information through existing fiber networks. Willow advances the computing side: protecting information inside a processor and showing that its circuits can reach a scale that challenges classical simulation. Applying those capabilities to chemistry, materials, or other useful calculations is the next target.
Sources and related information
Nature – Quantum error correction below the surface code threshold – 2024
The peer-reviewed Willow error-correction paper reports below-threshold surface-code memories, logical error suppression as code distance grows, real-time decoding, and the remaining barriers to large fault-tolerant computation.
Nature – Phase transitions in random circuit sampling – 2024
The peer-reviewed Sycamore study examines how noise changes RCS complexity and reports a 67-qubit, 32-cycle experiment beyond the capabilities of existing classical supercomputers.
Google – Meet Willow, our state-of-the-art quantum chip – 2024
Google’s Willow announcement provides the under-five-minute runtime, its 10^25-year Frontier estimate, the assumptions surrounding that comparison, and the company’s description of RCS as a system benchmark.
Google Research – How to compare a noisy quantum processor to a classical computer – 2023
Google’s technical explanation documents the earlier 47-year estimate for a Sycamore RCS experiment and explains how circuit complexity, fidelity, and usefulness are separate engineering dimensions.
Nature – Author Correction: Quantum error correction below the surface code threshold – 2026
The author correction identifies amended labels in Fig. 3a. The corrected labels do not alter the core Willow findings summarized in this article.
SciencePost – Google quantum computer completes a 47-year calculation in seconds – 2026
The SciencePost article prompted this update. Its 47-year comparison was checked against Google’s 2023 explanation and the peer-reviewed 2024 Sycamore paper before being placed in the timeline.
