Google Quantum AI

Google Quantum AI is Google's quantum-computing research group, which builds gate-model processors from superconducting qubits. It is known for the 2019 Sycamore experiment that claimed Quantum supremacy, a claim contested by IBM, and for the 2024 Willow chip, which demonstrated a key error-correction milestone.

#Background

Google's quantum effort operates as a dedicated research group that designs its own superconducting chips along with the cryogenic and control systems that run them. Its public work has concentrated on two goals: demonstrating computational tasks believed to be beyond classical reach, and showing that quantum error correction improves as the code is scaled up. The two Sycamore-era supremacy experiments served the first goal, while the Willow error-correction results serve the second, which is the one more directly connected to building a useful machine.

#Sycamore and the 2019 supremacy claim

In 2019 Google reported that its 53-qubit Sycamore processor sampled the output of random quantum circuits in about 200 seconds, a task the team estimated would take the leading classical supercomputer roughly 10,000 years (Arute et al. 2019). The paper framed this as the first demonstration of quantum supremacy, meaning a quantum machine performing a specific task beyond the reach of classical computers in feasible time.

IBM contested the classical estimate almost immediately. Researchers at IBM argued that a classical supercomputer, using secondary storage and a better simulation strategy, could reproduce the sampling in about 2.5 days rather than 10,000 years (Pednault et al. 2019). Later classical algorithms narrowed the gap further. The episode is now a standard illustration that supremacy claims depend on the best available classical method, which keeps improving, and the field increasingly prefers the more measured phrase Quantum advantage. Both facts hold together: Sycamore was a genuine milestone in random-circuit sampling, and the specific speedup figure was disputed and later reduced.

#Willow and below-threshold error correction

Google's more consequential result came in December 2024 with Willow, a 105-qubit superconducting chip. Using the Surface code, the team encoded a single logical qubit at increasing code distances of 3, 5, and 7 and showed that each increase in distance roughly halved the logical error rate, with the encoded qubit outliving the best individual physical qubit on the chip (Google 2024; Nature 2024).

This "below-threshold" scaling is the property fault tolerance depends on: it means adding more physical qubits to the code makes the logical qubit better rather than worse. The experiment also required decoding errors quickly enough to keep up with the hardware, so real-time error decoding was part of the result. It was demonstrated for a quantum memory at small scale, not for a full computation, and it does not by itself produce a useful algorithm. It is nonetheless widely regarded as one of the field's most important experimental gates, because sub-threshold operation had long been a theoretical requirement without a clean hardware demonstration. The chip's name, Willow, refers to the specific 105-qubit device rather than to a product line.

#Benchmarks and scope

Alongside these results Google has continued to run random-circuit-sampling benchmarks at larger scale, again framed as Quantum advantage rather than practical computation. As with the 2019 experiment, such benchmarks are contrived to be hard for classical machines and have no direct application. As of early 2026, Google has not reported a fault-tolerant machine or an advantage on a commercially or cryptographically meaningful problem.

#Relation to cryptography

Google's processors are universal gate-model machines, so a sufficiently large and error-corrected successor could in principle run Shor's algorithm. Willow is many orders of magnitude too small for that. The Sycamore supremacy result, often misread as a cryptographic breakthrough, concerns sampling and says nothing about when RSA or elliptic-curve keys become vulnerable. The Willow error-correction milestone is more relevant to the long-run threat, because scaling error correction is the central obstacle between today's machines and a cryptographically relevant quantum computer.