Google Quantum AI announced that it has practically implemented a technique known as “magic state cultivation” on its Willow quantum processor, achieving an approximately 40-fold reduction in errors compared with uncultivated states. The results were published in Science, in a step the company says addresses one of the most difficult resource bottlenecks hindering the expansion of quantum computing toward practical applications.
Why are magic states a bottleneck?
General-purpose quantum computing requires a specific set of operations to run useful programs that exceed the capabilities of classical computing. To perform these operations, special quantum states called “magic states” are created and then used in error-corrected operations. However, creating high-quality, error-corrected states consumes enormous resources; estimates cited by Google indicate that as much as 95% of a quantum computer’s capacity may be devoted to producing them rather than performing the computations themselves.
The problem is compounded because a magic state is consumed when used and cannot be reused, making it a scarce resource in the quantum system.
How did the cultivation technique work?
The idea, developed by Google Quantum AI researchers based on earlier work, involves introducing a noisy state into a small region of the chip and then carrying out a series of coordinated measurements to detect and discard faulty attempts. In this way, a pure, low-error quantum state is cultivated while simultaneously being adapted to a larger encoding that can be used inside the computer.
The main challenge was turning the theoretical idea into a system that operated on real hardware. During testing, the team discovered that drift in the electrical equipment and changes in temperature produced a type of noise that affected the validity of the measurement method used to evaluate the results. After simulations, consistency checks, and a review of the assumptions, the researchers added controls to reduce the impact of the noise and developed a new measurement technique that uses the cultivation process itself as a calibration tool.
Measurement results and what changed in practice
On the Willow processor, the protocol achieved a 40-fold reduction in errors compared with uncultivated states and produced T states with a logical fidelity of 99.99%, a result Google described as a new record level in quantum measurement.
The team also demonstrated the ability to transition between two types of quantum error correction: “color codes,” which are suitable for the cultivation process, and “surface codes,” a more established method. This included transferring states from a color code with distance 3 to a surface code with distance 5, while the purified states remained intact and were transferred within the hardware.
What has not yet been settled?
The result does not prove that practical quantum computing is now available. Google acknowledges that magic state cultivation needs to be demonstrated at a larger scale to become realistic for useful quantum computing, and other engineering and technical challenges remain. The achievement’s actual value is that it moves a central idea from the realm of theory to a successful experiment on a real processor and provides a potential path toward reducing the amount of resources required to build an error-corrected quantum computer.