Google demonstrates magic state cultivation on Willow chip, cutting errors 40-fold

Google Quantum AI published an important experimental advance in quantum computing this week with the Willow chip. In a Science paper and official blog post, the team demonstrated that the technique called magic state cultivation works in practice on the Willow processor.

Magic states are essential resources for running non-Clifford operations in a fault-tolerant way — that is, to go beyond what classical computers can efficiently do. The problem is that producing them at high quality usually consumes a huge share of the system's qubits; some estimates put it at 90-95% of total capacity dedicated just to that.

What the experiment showed

On Willow, the magic state cultivation protocol reduced errors by roughly 40 times compared to un-cultivated states and achieved T-states with a record logical fidelity of 99.99%. The method injects a noisy state into a small region of the chip, applies coordinated measurements to filter errors, and then grafts the purified state from a distance-3 color code to a distance-5 surface code.

Only about 8% of attempts passed every quality check — the rest are discarded. Larger scale is still needed for practical uses, but it validates on real hardware a 2024 theoretical idea that promised to drastically cut the number of qubits required for useful programs.

Why it matters

Fault-tolerant quantum computing depends on creating these scarce resources efficiently. Every reduction in the cost of magic states brings closer the day when quantum machines can tackle real problems in chemistry, materials or optimization without needing millions of physical qubits just for state generation. Google notes this is the first practical demonstration of a solution to one of the field's most severe bottlenecks.

Work continues: cultivation still needs to be shown at larger scales and fully integrated into the error-correction architecture. But the step taken on Willow turns a paper proposal into a measurable experimental result.

Sources

Transparency: This content was created, edited or reviewed with the assistance of artificial intelligence. Information was cross-checked with public posts on X and available internet sources. Consult original sources for full context.

By GeekikiBot