Google says magic state cultivation cuts errors about 40x on Willow chip
Google Quantum AI says the technique eases a long-known resource bottleneck, though it still has to be shown at a larger scale.
AI illustrationGoogle Quantum AI said on October 15 that it has shown a technique called magic state cultivation working on its Willow processor, cutting errors about 40x compared with running without it. The company said the work has now been published.
Magic states are special quantum states that help a machine carry out the set of operations needed for universal quantum computing, the kind able to run useful programs beyond what classical computers can do. Researchers have seen their production as a difficult bottleneck since 2004. According to Google, making high-quality, error-corrected magic states takes so many resources that some estimates suggest up to 95% of a quantum computer's capacity might need to go to them. Each magic state can also be used only once.
The cultivation method was developed in 2024 by Google Quantum AI researchers, who refined an idea from earlier literature. It is meant to act like a filter: a noisy state is placed in a small region of the chip, and a sequence of coordinated measurements checks for errors and discards failed attempts. The aim is to grow a very pure, low-error state that can at the same time be adapted into a larger encoding. Google said that, if it worked on hardware, the approach could substantially reduce the size of quantum computer needed for practical tasks.
On Willow, the protocol achieved a 40-fold reduction in errors relative to states that were not cultivated, and produced T-states with what the company called a record logical fidelity of 99.99%. Google described the result as the first demonstration of a practical solution to one of the most severe resource bottlenecks in the field.
Getting there required further engineering. During early tests, the team found that lab noise from electrical equipment drift and temperature variation was undermining its usual benchmark. It added extra control to reduce that noise and devised a new benchmarking method that uses cultivation to measure itself.
The researchers also moved states from a distance-3 color code to a distance-5 surface code, two forms of quantum error correction, showing that purified states can survive and transfer inside the hardware.
Google said magic state cultivation still has to be shown at a larger scale before it is realistic for practical quantum computing, and that plenty of technical challenges remain.