Google’s Willow Quantum Chip: A Step Forward or Just Quantum Noise?
February 11, 2025
By Margie Christ
Google’s recent announcement of their new “Willow” quantum computing chip has created a buzz that is hard to ignore. Sorting the pure hype from the actual achievements can be difficult, especially in the highly technical world of quantum computing. The following piece hopes to help clear the quantum fog to identify what is a step forward for Willow, and what is pure quantum noise.
The main achievement of the Willow chip is efficient quantum error correction (QEC). QEC is important because qubits are profoundly impacted by their environments, and typically only stay in the state we need for a very short period of time. This makes them highly susceptible to error. QEC helps qubits behave in a desirable manner for solving computations and generally performing tasks on a quantum computer. To illustrate, imagine an important document, such as a business contract. If only one copy of the contract exists, any accident like spilling coffee or sending it through the wash would be unfortunate. Keeping multiple redundant copies of important documents can be a safeguard against losing information. QEC often functions similarly by redundantly encoding information across multiple qubits. This gives rise to the difference between “physical” qubits (the raw number of qubits on a quantum chip, or each individual copy of the document in our analogy) and “logical” qubits (the functional number of qubits that can be used for computations; each logical qubit is composed of a collection of physical qubits, or the sum total of all of our redundant document copies). In practice, this makes it more difficult to scale up quantum computers, as the effective number of qubits ends up being less than the actual number of qubits on the chip. This is a negative spiral: fewer qubits means smaller computations, which means solving smaller problems, which means answering smaller questions.
Another problem with QEC is that it has, historically, gotten more and more difficult to perform as more qubits are added to a chip – making the error rate increase with the addition of more qubits, and preventing scalability. Combining multiple physical qubits, each with their own individual error rates, to perform QEC can serve to compound those individual error rates, making things worse instead of better.
This is why Google's Willow chip is a significant step forward in achieving sustainable QEC: When adding more qubits, the Willow architecture achieves a decrease in error rate, exactly the phenomenon necessary for scalable QEC. The researchers compare the base error rate of a single qubit (around 3 × 10⁻³ errors per cycle) to the error rate of single logical qubit on the Willow chip composed of a 7x7 grid of physical qubits (around 1.4 × 10⁻³ errors per cycle). Despite adding qubits, the 7x7 grid has an error rate half that of a single qubit!
The researchers cite improved qubit production (better fabrication techniques, improved participation ratio engineering, and improved circuit parameter optimization) as one contribution to this result, as well as the improved architecture of the Willow chip. They also saw longer coherence times, with the 7x7 (or distance-7) qubit having more than double the lifespan (291 ± 6 microseconds) of its individual component qubits (median time is 85±7 microseconds). Longer coherence time gives more time to perform calculations, allowing for larger problems to be considered.
The researchers note that though this is an excellent step forward, much work remains. The researchers state that there are “orders of magnitude remain between present logical error rates and the requirements for practical quantum computation”, say the researchers, as an error rate of 10⁻⁶ is considered necessary for a fully functional quantum computer. These advancements are promising, and constitute a big step towards functional, scalable gate-model quantum computing. However, there is still work that needs to be done to make this goal a reality (the researchers already identify possible improvements to be implemented in future work).
The Willow chip improvements promise scalability in theory, but in practice will be extremely resource intensive – based on their projections, achieving the famed 10⁻⁶error rate would require a single logical qubit be composed of 1457 physical qubits; a number that far exceeds the size of current gate-model chips. Furthermore, there are also limitations on the classical computing side: quantum computations require decoding, and this decoding must also increase in performance with increases in performance of the quantum chip.
The Willow chip is an excellent proof of principle, and has achieved extremely promising results. It is not, however, the final word in quantum computing – more work is still necessary to take these results to the next level of usability! According to Google’s roadmap for achieving fully functional quantum computing, they have achieved “Milestone 2” of their 6-milestone plot (see diagram at bottom of page: https://blog.google/technology/research/google-willow-quantum-chip/)
In terms of impact on the Canadian quantum technology landscape, many of the bigger players in the Canadian QC ecosystem are not using superconducting qubits (i.e. Xanadu uses photonic/optical qubits, Photonic uses the same, etc.), so it’s hard to know what the impact may be. Notably, however, Anyon does use superconducting qubits, so there may be direct impact in some areas. Nord Quantique likely has the most similar technology, as they also develop error-corrected superconducting qubits. It will be interesting to see the impact – Nord Quantique is focused on qubit-efficient QEC, so Google’s method of using huge numbers of physical qubits for error correction on the Willow chip is very different from the research direction being pursued by NQ.
While the Willow chip does not solve all the challenges of quantum computing, it represents a major step forward. It is a major achievement for QEC, and it helps set a new direction for future progress in the field, hopefully allowing for the construction of bigger and better quantum chips.