Quantum Companies Database
Google (Google Quantum AI)
Google Quantum AI builds superconducting quantum processors, including the 105-qubit Willow chip, which showed that error rates can fall as more qubits are added. In March 2026 the lab added a second track based on neutral atoms.
Founded
Quantum AI Lab launched in 2013
CEO
Sundar Pichai (Alphabet and Google CEO); Hartmut Neven (founder and lead of Google Quantum AI)
Headquarters
Mountain View, California, USA (quantum hardware lab in Santa Barbara, California)
Technology
Superconducting qubits (neutral atom program added in 2026)
Funding / Status
Part of Alphabet Inc. (NASDAQ: GOOGL); funded from Alphabet's research budget
Website
quantumai.google
Overview
Google Quantum AI is the quantum computing research group inside Google, part of Alphabet. Its main lab in Santa Barbara, California designs and builds its own superconducting quantum chips, and a newer team in Boulder, Colorado is working on neutral atom hardware. Unlike IBM, Google has not made its processors generally available on a public cloud. Access is limited to Google's own researchers and to selected outside research teams.
The group has produced some of the field's most cited results: the 2019 quantum supremacy experiment on Sycamore, the 2024 demonstration on Willow that error correction improves as a chip scales up, and the 2025 Quantum Echoes experiment, which Google described as the first verifiable quantum advantage. Its stated goal is a large-scale, error-corrected quantum computer, and Google has said it expects commercially relevant machines by the end of this decade.
History and background
Google's quantum work began around 2012 and 2013 with the Quantum Artificial Intelligence Lab, set up with NASA and the Universities Space Research Association. The lab first experimented with quantum annealing hardware from D-Wave. In 2014 Google decided to build its own gate-based processors and brought in John Martinis and his superconducting-qubit group from the University of California, Santa Barbara.
That decision produced the 72-qubit Bristlecone chip in 2018 and the 53-qubit Sycamore in 2019. In October 2025 the Nobel Prize in Physics went to John Clarke, Michel Devoret, and John Martinis for their 1980s experiments on superconducting circuits, work that underpins today's superconducting qubits. Devoret is Chief Scientist for Quantum Hardware at Google Quantum AI, and Martinis is a former leader of Google's hardware effort.
How the technology works
Google's chips use superconducting transmon qubits arranged on a two-dimensional grid, where each qubit interacts with its four nearest neighbors. Willow has 105 qubits. According to figures published alongside the Quantum Echoes result, single-qubit gates reach about 99.97% fidelity, entangling gates about 99.88%, and readout about 99.5%.
The most important idea in Google's roadmap is the logical qubit. A single physical qubit is too noisy for long computations, so error correction spreads one logical qubit across many physical ones. In December 2024, Willow showed that when Google increased the size of its surface code, from a 3 by 3 grid of qubits to 5 by 5 and then 7 by 7, the logical error rate fell at each step, by roughly half. This below-threshold behavior is what theory says is needed for error correction to scale.
Google's public roadmap has six milestones. Willow sits between the second milestone, a prototype error-corrected qubit, and the third, a long-lived logical qubit with very low error rates. The end goal is a machine with about a million physical qubits. In March 2026 Google added neutral atom hardware as a second track: superconducting chips are fast and good at deep circuits, while neutral atom arrays can reach many thousands of qubits with flexible connectivity but run more slowly.
Products and platforms
Willow
Google's 105-qubit superconducting processor, announced in December 2024. Google reported that a benchmark called random circuit sampling took Willow only minutes, a task it estimated would be impractical for a leading supercomputer. The more meaningful result was the below-threshold error correction. Willow is not on a public cloud; Google opened an early-access program for selected research proposals in 2026.
Quantum Echoes
An algorithm run on Willow and published in Nature in October 2025. It sends a signal into the quantum system, perturbs one qubit, and then reverses the evolution to listen for an echo. Google reported it ran about 13,000 times faster than the best classical method it tested, and called the result verifiable because another quantum computer of similar quality should be able to reproduce it. A proof-of-principle experiment with UC Berkeley applied the technique to two molecules using NMR data.
Sycamore
The 53-qubit processor behind the 2019 quantum supremacy claim. Later classical simulations narrowed the gap Google originally described, but the experiment remains a landmark in the field. See the Sycamore entry in our Hardware Database.
Cirq
Google's open-source Python framework for writing, simulating, and running quantum circuits. Cirq is used mainly by researchers and can be run on your own computer without access to Google's hardware.
Neutral atom program
Announced on March 24, 2026 and led by physicist Adam Kaufman in Boulder, Colorado. The program focuses on error correction for atom arrays, simulation of hardware designs, and experimental hardware. Google also continues to work with QuEra, a neutral atom company in which it is an investor.
Processors in our Hardware Database
- Google Sycamore — 53 qubits, released 2019
- Google Willow — 105 qubits, released 2024
Key milestones
| When | What happened |
|---|---|
| 2013 | The Quantum Artificial Intelligence Lab launches with NASA and USRA. |
| 2014 | Google announces plans to build its own superconducting processors and recruits John Martinis's team. |
| 2018 | Bristlecone, a 72-qubit chip, is unveiled. |
| 2019 | Sycamore (53 qubits) completes a random circuit sampling task that Google says demonstrates quantum supremacy. |
| 2023 | Google shows a distance-5 surface code performing slightly better than a distance-3 code on Sycamore-class hardware. |
| December 2024 | Willow (105 qubits) demonstrates below-threshold error correction. |
| October 2025 | Quantum Echoes is published in Nature; the Nobel Prize in Physics is shared by Michel Devoret and John Martinis with John Clarke. |
| March 2026 | Google announces a neutral atom hardware program and opens the Willow Early Access Program for research proposals. |
Strengths and challenges
Strengths
- Leading published error-correction results, including below-threshold scaling on Willow.
- Deep research talent, including a Nobel laureate as Chief Scientist for hardware.
- Large computing resources and AI expertise for simulating designs and decoding errors.
- A clearly staged roadmap with defined milestones.
Challenges and open questions
- Willow is not publicly accessible, so outside researchers cannot easily verify or build on its results.
- The long-lived logical qubit milestone, and everything after it, has not been reached yet.
- The 2019 supremacy claim was later challenged by improved classical simulation, so new advantage claims invite scrutiny.
- Running two hardware tracks, superconducting and neutral atom, adds cost and complexity, and the neutral atom effort is at an early stage.
Why it matters
Google's results shape how the whole field thinks about error correction. The Willow experiment was widely seen as the first clear evidence that adding more physical qubits can make a logical qubit better rather than worse, which is the central bet behind every fault-tolerant roadmap.
There is also a security angle. In 2025 a Google researcher published an analysis suggesting that RSA-2048 encryption could be broken with fewer than a million noisy qubits in under a week, far below earlier estimates. That is still far beyond today's hardware, but it is one reason governments and companies are planning migrations to post-quantum cryptography.
Latest news
- Google Quantum AI launches a neutral atom hardware program led from Boulder, Colorado — March 2026
- Willow Early Access Program opens for outside research proposals (deadline May 15, selections by July 1) — March 2026
- Michel Devoret, Google Quantum AI's Chief Scientist for hardware, shares the Nobel Prize in Physics — October 2025
- Google publishes Quantum Echoes on Willow in Nature, reporting a 13,000x speedup over classical methods — October 2025
Frequently asked questions
What is Google Willow?
Willow is Google's 105-qubit superconducting quantum processor, announced in December 2024. Its key result was showing that a logical qubit's error rate falls as the error-correcting code is made larger, which is a requirement for building fault-tolerant machines.
Can I use Google's quantum computer?
Not directly. Willow is not on a public cloud, though Google has run an early-access program for selected research teams. You can use Google's open-source Cirq framework to write and simulate circuits on your own computer.
Did Google really achieve quantum supremacy?
Google claimed it in 2019 with Sycamore. Other researchers later showed that improved classical algorithms could simulate the task much faster than Google originally estimated, so the claim remains debated. The 2025 Quantum Echoes result was framed differently, as a verifiable advantage on a physically meaningful problem.
What is a logical qubit?
A logical qubit is a more reliable qubit built from many physical qubits using error correction. If the physical qubits are good enough, adding more of them makes the logical qubit's error rate drop, which is what Willow demonstrated.
Why did Google start a neutral atom program?
Google says the two technologies complement each other. Superconducting chips run very fast and handle deep circuits well, while neutral atom arrays scale to many thousands of qubits with flexible connectivity. Pursuing both lets the lab hedge its bets and share research between them.
This profile is written for educational purposes and summarizes public announcements, company filings, and published research. Last reviewed: September 30, 2026. It is not investment advice. Check the company's official channels for the latest figures and claims.