How Google Built and Tested the Willow Quantum Processor
Google’s 105-qubit Willow processor combined improved superconducting hardware with below-threshold quantum error-correction results and a fast random-circuit-sampling benchmark. The milestones advance fault-tolerant research but do not make Willow a general-purpose commercial computer.
Timeline
- Before 2024: Google developed Willow’s superconducting processor, fabrication process, controls and decoding systems.
- December 9, 2024: Google announced Willow and the below-threshold error-correction paper was published online in Nature.
- Next research stage: Google identified useful beyond-classical computation as a remaining milestone rather than a completed capability.
Willow is Google Quantum AI’s superconducting quantum processor, announced in December 2024 with 105 physical qubits. Its importance comes from system-level experiments rather than the qubit count alone. Google highlighted two results: a quantum error-correction experiment in which logical errors fell as the encoded system grew, and a random-circuit-sampling benchmark completed in under five minutes. Those are separate achievements with different meanings. [1][2]
Quantum information is fragile because qubits are affected by heat, electromagnetic interference, imperfect controls and other noise. Error correction spreads one logical unit of information across multiple physical qubits and repeatedly measures clues about errors without directly reading the protected state. This only becomes scalable when the physical system operates below a threshold at which adding a larger error-correcting code suppresses logical errors instead of adding more failures. [2][3]
The Nature paper reports below-threshold surface-code memories on Willow processors. Its larger distance-7 memory used 101 qubits and had a logical error rate per correction cycle that fell by more than half when code distance increased by two. The researchers also reported “beyond breakeven” behavior: the logical memory lasted longer than its best constituent physical qubit. These results concern protected quantum memory, not a completed fault-tolerant application. [1][2]
The team also integrated real-time decoding with a distance-5 code. A decoder interprets error-syndrome measurements and decides which correction should be tracked while the computation continues. The paper reports an average decoder latency of 63 microseconds in that experiment and operation for up to one million cycles. Real-time performance matters because corrections that arrive too late cannot protect a running quantum computation. [2]
Willow’s other headline result used random circuit sampling, a benchmark designed to be extremely hard to reproduce classically. Google reported that Willow finished the selected computation in under five minutes and estimated that a leading classical supercomputer would require about 10^25 years under its comparison assumptions. The estimate is benchmark-specific, depends on classical algorithms and hardware assumptions, and is not equivalent to speeding up ordinary software by that factor. [1]
Building Willow required more than designing a circuit diagram. Google says it fabricated the chip in its Santa Barbara facility using superconducting integrated circuits, then combined qubit fabrication, control gates, reset, readout, wiring, shielding and very low-temperature refrigeration. Its lab description explains that superconducting qubits operate in a dilution refrigerator and need careful packaging to reduce radio, thermal and electromagnetic disturbances. [1][3]
Willow therefore demonstrates progress toward scalable error correction and strong performance on a specialized benchmark. It does not yet provide a known commercially useful computation beyond classical reach. Google explicitly identified that combination as the next challenge: an algorithm that classical computers cannot practically reproduce and that also solves a real-world problem. Claims that Willow has already transformed medicine, energy or everyday computing go beyond the published evidence. [1][2]
Sources
- Google — Meet Willow, our state-of-the-art quantum chip
- Nature — Quantum error correction below the surface code threshold
- Google — Go inside the Google Quantum AI lab