Google’s Willow: The Quantum Chip That Solves in 5 Minutes What’s Impossible to Calculate

Google Quantum AI has unveiled Willow, a quantum chip that represents a major breakthrough in the race for quantum computing. This 105-qubit processor has accomplished the impossible: solving in less than five minutes a calculation that would require 10 septillion years for the most powerful supercomputers in the world. What makes Willow so exceptional? How does quantum error correction work? What concrete applications can be envisioned? Breaking down a historic technological breakthrough.

Google Quantum AI’s Willow quantum chip — Source: Google

What is Willow and why is this announcement historic?

Willow is the new quantum processor developed by Google Quantum AI, the team founded by Hartmut Neven in 2012 with the ambition of building a useful large-scale quantum computer. This superconducting chip contains 105 qubits and succeeds the Sycamore processor that made waves in 2019.

The announcement of Willow marks two simultaneous major accomplishments. First, the processor has demonstrated the ability to exponentially reduce errors as the number of qubits increases, a challenge researchers have been trying to solve for nearly 30 years. Second, Willow executed a reference benchmark in less than five minutes while the Frontier supercomputer, one of the fastest in the world, would require 10^25 years to accomplish the same task.

To put this number in perspective: 10 septillion years (or a 1 followed by 25 zeros) far exceeds the age of the universe, estimated at approximately 13.8 billion years. This spectacular demonstration testifies to the potential of quantum systems to accomplish operations fundamentally inaccessible to classical computing.

Quantum error correction: the Holy Grail finally achieved

The historic challenge of qubit stability

Qubits, the fundamental units of quantum computing, have a problematic characteristic: they rapidly exchange information with their environment, compromising ongoing calculations. Traditionally, increasing the number of qubits proportionally amplified errors, making the system unusable.

Quantum error correction (QEC) aims to protect quantum information from these disturbances. The concept, introduced by Peter Shor in 1995, relies on grouping multiple physical qubits to form a more robust “logical qubit.” However, demonstrating that this approach actually works in practice has remained a major challenge.

Willow crosses the critical threshold

Google’s team tested increasing qubit configurations: 3×3 grids, then 5×5, and finally 7×7 encoded qubits. With each scale increase, the error rate was cut in half. This exponential reduction of errors with increasing number of qubits constitutes the demonstration of “below threshold” operation, a historic accomplishment in the field.

Michael Newman, a researcher at Google, explains the importance of this breakthrough: groupings of qubits in surface codes now allow errors to be suppressed exponentially fast as the system scales. The teams achieved an error rate of approximately 0.14% per cycle, a major improvement compared to previous generations.

Real-time correction: a first

Willow also represents one of the first convincing examples of real-time error correction on a superconducting quantum system. This capability proves crucial for any practical application: if errors cannot be corrected quickly enough, they ruin the calculation before completion.

The “beyond break-even” demonstration proves that qubit groupings possess a lifespan superior to individual physical qubits, an unmistakable sign that error correction improves the system overall.

Willow chip technical specifications

Architecture and performance

Willow contains 105 superconducting transmon qubits, arranged in a square grid. The chip was manufactured at Google’s state-of-the-art facilities in Santa Barbara, one of the rare factories in the world built specifically for quantum processor production.

The T1 coherence time of the qubits, measuring the duration a qubit can retain an excitation, now approaches 100 microseconds. This performance represents a 5-fold improvement compared to the previous Sycamore generation.

Improvements over Sycamore

The team led by Julian Kelly, director of quantum hardware at Google, emphasizes that all system components must function simultaneously in an optimal manner. Improvements concern manufacturing as much as circuit parameter optimization and participation ratio engineering.

The RCS benchmark: measuring quantum supremacy

What is Random Circuit Sampling?

Random Circuit Sampling (RCS) constitutes the reference benchmark for evaluating quantum performance. Pioneered by the Google team and now standard in the field, this test represents the most difficult task to simulate classically that a quantum computer can perform.

This benchmark verifies whether a quantum computer truly accomplishes something impossible for a classical computer. Any team developing a quantum processor should first validate its ability to outperform classical machines on this test before claiming to solve more complex quantum problems.

Willow’s results in context

Willow’s performance on this benchmark exceeds all expectations. The calculation executed in less than five minutes would require the Frontier supercomputer approximately 10^25 years, even granting it impossible ideal conditions in practice, such as unlimited memory access with zero bandwidth latency.

Google used this same benchmark to announce its Sycamore results in October 2019, and again in October 2024. The growing gap between quantum and classical capabilities follows a doubly exponential progression, suggesting that quantum processors will continue to outpace their classical counterparts.

Future applications of quantum computing

Drug discovery and molecular chemistry

Molecular simulations represent one of the most promising applications of quantum computing. Quantum systems can simulate atomic interactions with a fidelity impossible to achieve with classical computers, accelerating the discovery of new drugs and understanding of complex chemical reactions.

Battery design and materials

The optimization of batteries for electric vehicles and the discovery of new materials would benefit considerably from quantum simulation capabilities. Researchers could explore atomic configurations impossible to test experimentally, identifying optimal solutions more quickly.

Fusion energy and climate challenges

Hartmut Neven mentions the solving of currently unsolvable problems such as optimizing nuclear fusion or combating climate change. These applications will require quantum computers far more powerful than Willow, but current advances trace the path toward these objectives.

Artificial intelligence and machine learning

Artificial intelligence could significantly benefit from quantum computing for training certain learning architectures, gathering training data inaccessible to classical machines, and modeling systems where quantum effects play an important role.

Current limitations and perspectives

A research prototype, not yet commercial

Despite these impressive advances, Willow remains a research prototype in the NISQ (Noisy Intermediate-Scale Quantum) era. The logical error rates reported, while impressive, remain several orders of magnitude above the 10^-6 levels deemed necessary for executing large-scale useful quantum algorithms.

Demonstrations have so far been limited to quantum memory and preservation of logical qubits, without yet showing below-threshold performance for logical gate operations required for fault-tolerant computation.

The cryptography question

The announcement of Willow has revived concerns about the security of cryptocurrencies and current encryption systems. However, Google specifies that Willow remains incapable of breaking modern cryptography. Experts estimate that a quantum computer possessing this capability remains 5, 10, or even 15 years away.

NIST (National Institute of Standards and Technology) has already published several “quantum-safe” algorithms resistant to attacks from future quantum computers, and a deployment timeline is being developed for governments and enterprises.

Roadmap toward useful quantum computing

Google published a detailed roadmap in 2020, and the company affirms it is staying on schedule. The next major milestone, called “Milestone 6,” plans for the arrival of large-scale quantum machines toward the end of the decade.

The intermediate objective consists of demonstrating the first “useful and beyond-classical” calculation: a task simultaneously beyond the reach of classical computers AND presenting utility for concrete applications with commercial relevance.

Google facing quantum competition

Google is not alone in the race for quantum computing. IBM is also developing superconducting systems, although its qubits exhibit shorter lifespans than those of Willow. Microsoft and Quantinuum announced in September 2024 promising results using laser-trapped qubits, with 12 logical qubits displaying an error rate of two per thousand.

Competition stimulates innovation across the entire sector. Private and public investments in quantum computing globally totaled approximately 20 billion dollars over the past five years, testifying to the strategic importance accorded to this technology.

What to take away from Willow

Google’s Willow quantum chip represents a significant advance toward practical quantum computing. The demonstration of error correction below the critical threshold proves that quantum systems can indeed improve as they scale up, a fundamental prerequisite for building useful machines.

However, concrete commercial applications remain several years away. Willow constitutes a major proof of concept, but substantial hardware improvements and much larger qubit architectures will be necessary before solving industrially relevant problems.

For Hartmut Neven, the choice to leave the nascent AI field to dedicate himself to quantum computing remains justified: both technologies will be the most transformative of our era, and advanced AI will benefit significantly from access to quantum computation. This is why the laboratory is named “Quantum AI.”

Sources: Google Blog — Meet Willow | Scientific American | HPCwire