Google Willow: The 105-Qubit Quantum Chip That Just Made Supercomputers Obsolete?
Google’s Willow quantum chip delivered a result that sounds almost impossible: a benchmark completed in minutes that Google estimated would take a leading classical supercomputer an astronomically long time. But does that really mean conventional supercomputers are obsolete? The answer is more interesting — and more important — than the headline.
- What Google Willow actually is
- Why 105 qubits matters — and why qubit count alone is not enough
- How quantum computers differ from classical supercomputers
- Why Willow’s error-correction result is arguably the bigger breakthrough
- What the “10 septillion years” claim really means
- What happened with Google’s later Quantum Echoes breakthrough
- What Willow could mean for AI, medicine, chemistry, batteries and materials
- Why supercomputers are not disappearing
- The biggest technical challenges still standing between Willow and a useful fault-tolerant quantum computer
Quantum computing has spent decades living in an unusual place between revolutionary science and futuristic promise. Researchers have repeatedly demonstrated that quantum systems can behave in ways classical computers cannot efficiently reproduce. Yet turning those laboratory demonstrations into reliable, useful machines has been extremely difficult.
Google’s Willow processor is important because it attacks one of the hardest problems in the entire field: quantum errors. The chip contains 105 physical qubits, but the headline number is not the main story. The deeper story is that Google demonstrated a form of quantum error correction in which increasing the size of the error-correcting code actually reduced the logical error rate. That is exactly the direction researchers need if they want to build large, fault-tolerant quantum computers.
.jpg)
What Is Google Willow?
Willow is a superconducting quantum processor developed by Google Quantum AI. Google announced the chip in December 2024 and described it as a major step toward useful, large-scale quantum computing.
The processor contains 105 superconducting qubits. These are not miniature versions of the bits used in your laptop or phone. A classical bit is represented as either 0 or 1. A quantum bit, or qubit, is a quantum system that can be prepared in a combination of states and manipulated through quantum operations.
Willow’s qubits are based on superconducting circuits. Such systems have to be operated under extremely cold conditions because the quantum behavior being exploited is delicate. The processor is therefore part of a much larger system involving cryogenic cooling, microwave control electronics, readout hardware, calibration systems and software.
This is an important point: the quantum chip is not a standalone replacement for a supercomputer. A useful quantum computer is a complete system. The processor sits inside an environment designed to protect, control and measure quantum states with extraordinary precision.
Why 105 Qubits Is More Complicated Than “105 Bits”
It is tempting to look at the number 105 and conclude that a modern CPU or GPU with billions of transistors should be vastly more powerful. That comparison misses the fundamental difference between classical and quantum information.
Quantum computation uses concepts such as superposition, entanglement and interference. A quantum circuit manipulates amplitudes associated with possible computational states. With many qubits, the mathematical state space grows exponentially. For example, 10 classical bits can represent one of 1,024 possible binary states at a time. A 10-qubit quantum system has a state described across a 1,024-dimensional computational basis.
That does not mean a quantum computer simply “tries every answer at once” and then magically reads all of them. Measurement gives limited information. The real power comes from designing quantum operations so that unwanted possibilities interfere destructively while useful information is amplified.
This is why quantum algorithm design is so important. A quantum processor can be spectacularly fast for a particular mathematical structure while being completely unsuitable for an ordinary workload such as editing a video, running a web server, compiling a typical application or browsing the internet.
Quantum Computer vs Supercomputer: They Are Built for Different Jobs
A classical supercomputer is essentially a huge collection of conventional computing resources working together. Modern systems use enormous numbers of CPU and GPU cores, high-speed memory, networking fabrics and storage. They are extremely versatile.
A quantum computer is much more specialized. Its advantage appears only for algorithms that can exploit quantum mechanical behavior. This means the question should not be “Is Willow faster than a supercomputer?” The better question is:
That distinction matters because Google’s most famous Willow performance number comes from Random Circuit Sampling (RCS). RCS is designed to stress the ability of a quantum processor to generate and sample from complex quantum states. It is extremely difficult for classical computers to simulate as the system grows.
But RCS is not a practical workload like weather forecasting, protein design, banking, web search or video rendering. It is a benchmark intended to demonstrate a quantum processor operating beyond the practical reach of classical simulation.
The “10 Septillion Years” Claim Explained
This is the number that generated the biggest headlines.
Google said Willow completed a standard Random Circuit Sampling benchmark in under five minutes that it estimated would take one of today’s fastest classical supercomputers approximately 10 septillion years, or 1025 years, to perform the equivalent computation under the assumptions used in the comparison.
Ten septillion is an almost incomprehensible number:
10,000,000,000,000,000,000,000,000 years
The age of the universe is roughly 13.8 billion years. So the comparison is intentionally extreme. It illustrates how quickly the classical simulation cost of certain quantum systems can explode.
However, this number should not be interpreted as “Willow is 1025 times faster than a supercomputer at everything.” It is a comparison for a particular benchmark and a particular classical simulation strategy. Classical algorithms and hardware also improve over time, and benchmark methodology matters enormously.
The Bigger Breakthrough: Quantum Error Correction
If you ask many quantum-computing researchers what is more important about Willow than the giant benchmark number, the answer is likely to be error correction.
Qubits are fragile. Environmental noise, imperfect control pulses, unwanted interactions and measurement imperfections can introduce errors. In a classical computer, an individual bit can be protected with relatively straightforward error-detection and error-correction techniques. Quantum information is harder because quantum states cannot simply be copied in the classical sense.
A practical fault-tolerant quantum computer therefore needs a way to encode one reliable logical qubit across many imperfect physical qubits.
This creates a difficult engineering problem. You might think that adding more physical qubits should automatically make the computer better. But if every extra qubit introduces more opportunities for errors, scaling could actually make the system worse.
Willow demonstrated the opposite trend within its error-correction experiment: when Google increased the size of its surface-code memory, the logical error rate decreased. This is known as operating below threshold.
What Does “Below Threshold” Mean?
Imagine you are trying to protect a fragile message from noise. If your communication channel is too noisy, adding more error-correcting information may not help enough. But once the underlying error rate is low enough, increasing the error-correcting code can produce an increasingly reliable logical message.
Quantum error correction has a similar threshold behavior.
In the Willow experiments reported in Nature, Google Quantum AI demonstrated distance-5 and distance-7 surface-code memories. On the 105-qubit processor, the team implemented a distance-7 code using 101 qubits for the logical memory experiment and demonstrated a logical error rate that was suppressed as the code distance increased.
Nature reported that the larger logical memory had a logical error rate of about 0.143% per error-correction cycle in the reported experiment and that the distance-7 logical memory exceeded the lifetime of its best constituent physical qubit by a factor of about 2.4.
That “beyond breakeven” result is important. It means the error-corrected logical memory can preserve quantum information for longer than the individual physical qubit from which it is constructed. That is a critical sign that error correction is genuinely improving the system rather than merely adding complexity.
Why Error Correction Could Matter More Than Raw Qubit Count
Quantum-computing headlines often focus on qubit counts: 50 qubits, 100 qubits, 1,000 qubits, 10,000 qubits and eventually millions. But raw qubit count can be misleading.
Suppose one machine has 1,000 noisy physical qubits and another has 100 high-quality qubits with better control, connectivity and error correction. The second machine could be more useful for a particular algorithm.
The future of quantum computing is therefore not simply a race toward the biggest number printed on a chip specification sheet. Researchers need improvements in:
- Physical qubit fidelity
- Coherence time
- Two-qubit gate performance
- Readout accuracy
- Connectivity
- Calibration stability
- Real-time error decoding
- Logical-qubit lifetime
- Manufacturing yield
- Control-electronics scalability
Willow’s Superconducting Technology
Willow belongs to the superconducting-qubit approach to quantum computing. In simplified terms, these systems use engineered electrical circuits that behave quantum mechanically.
The circuits are cooled to extremely low temperatures so that thermal energy does not overwhelm the delicate quantum states. Microwave signals are then used to manipulate the qubits and readout operations are used to extract information.
This architecture has an important advantage: it can use fabrication techniques that resemble the highly developed world of semiconductor manufacturing. But there is also a major challenge. A useful system needs a huge amount of supporting infrastructure around the chip.
The cryogenic system, wiring, filters, amplifiers, microwave electronics and control software all have to work together. As the number of qubits grows, the engineering problem becomes enormous.
Willow’s Performance Is a System-Level Achievement
Google emphasized that the processor should not be judged only by the number of qubits. The company reported improvements across multiple system-level metrics. In its 2024 announcement, Google said Willow’s T1 times — a measure related to how long an excited quantum state survives — were approaching 100 microseconds, roughly five times better than the previous generation on that metric.
In later 2025 work on Quantum Echoes, Google reported very high gate fidelities across the 105-qubit array, including approximately 99.97% for single-qubit gates and 99.88% for entangling gates, along with approximately 99.5% readout fidelity.
These percentages may look close to perfect, but quantum algorithms can require huge numbers of operations. A tiny error probability repeated across millions or billions of operations can become a serious problem. That is why fault tolerance is essential.
From Random Circuit Sampling to Quantum Echoes
Willow’s story did not stop with the 2024 RCS benchmark.
In October 2025, Google announced a new result involving an algorithm it called Quantum Echoes. Google said the algorithm demonstrated what it described as the first verifiable quantum advantage on hardware, with the quantum computation running about 13,000 times faster than the best classical algorithm used for the comparison on a leading supercomputer.
This was a more interesting step because the goal was not simply to generate a hard-to-simulate random quantum state. Quantum Echoes was designed around an out-of-time-order correlator style computation and was aimed at extracting information about physical systems.
Google described the technique as a way of running a sequence of quantum operations, perturbing part of the system and then reversing the evolution to study the resulting “echo.” The idea can help researchers investigate how information and disturbances spread through quantum systems.
This matters because the long-term goal of quantum computing is not to win artificial benchmarks forever. The goal is to solve useful problems that matter to science and industry.
Could Willow Transform Medicine?
One of the biggest long-term promises of quantum computing is molecular simulation.
Molecules themselves obey quantum mechanics. Classical computers can simulate many chemical systems extremely well, but the computational cost can become enormous as the system becomes more complicated.
A sufficiently powerful fault-tolerant quantum computer could eventually simulate quantum systems more naturally than a classical computer. That could help researchers explore molecular structures, chemical reactions and interactions relevant to drug discovery.
It is important not to overstate the current situation. Willow is not currently a magical “drug discovery machine” that can instantly design medicines. The hardware and algorithms still have a long way to go. But the direction is significant: better error correction makes it more plausible that future quantum machines could run deep, reliable algorithms for chemistry.
Could Quantum Computing Improve Batteries?
Battery research is another frequently discussed application.
Designing a better battery involves understanding materials, ions, electrons, chemical reactions and interfaces. Many of these processes are quantum mechanical at their core.
A mature quantum computer could potentially help researchers model candidate materials and chemical mechanisms with greater accuracy. This might eventually accelerate the search for improved energy-storage materials, catalysts and other technologies.
Again, the word to emphasize is eventually. The current generation of quantum processors is still in the process of moving from laboratory demonstrations toward robust, application-level advantage.
What About Artificial Intelligence?
The relationship between quantum computing and AI is especially interesting.
Today’s AI boom is powered overwhelmingly by classical computing: GPUs, TPUs, CPUs, high-bandwidth memory and massive data-center infrastructure. Quantum computing is not replacing these systems.
Instead, quantum computing could eventually become a specialized accelerator for certain mathematical tasks. Potential research areas include optimization, sampling, quantum chemistry for training data, and problems where the underlying physical system is quantum mechanical.
Google has argued that quantum computation could complement advanced AI rather than compete with it. A future architecture could therefore look less like “quantum replaces GPUs” and more like “classical computers and quantum processors work together.”
So, Are Supercomputers Obsolete?
No.
At least not because of Willow.
Supercomputers remain extremely important. They run weather and climate simulations, computational fluid dynamics, nuclear physics calculations, astronomy workloads, AI training, engineering simulations, financial models, genomics pipelines and countless other applications.
Quantum computers are not general-purpose replacements for CPUs and GPUs. They are more comparable to a radically different computational architecture that may provide enormous advantages for a subset of problems.
In fact, the future is likely to be hybrid. A classical system may prepare a problem, send a specialized portion to a quantum processor, receive quantum measurements, and then use classical algorithms to interpret the results.
| Feature | Classical Supercomputer | Quantum Computer |
|---|---|---|
| Basic information unit | Bit | Qubit |
| General-purpose computing | Excellent | Limited / specialized |
| Quantum simulation | Can become extremely expensive | Potentially natural for suitable algorithms |
| Error behavior | Highly controllable at the logical level | Quantum states are fragile and require specialized correction |
| Cooling requirement | Depends on system design | Superconducting systems require extreme cryogenic cooling |
| Best use today | Broad scientific and commercial workloads | Quantum research and selected algorithms |
The Real Challenge: Building a Fault-Tolerant Quantum Computer
Willow should be viewed as a milestone on a much longer road.
The ultimate target is a fault-tolerant quantum computer with enough long-lived logical qubits to run algorithms containing enormous numbers of operations without the calculation being destroyed by accumulated errors.
That is a dramatically harder problem than building a 105-qubit processor.
The 105 physical qubits on Willow cannot simply be treated as 105 perfect computational units. A fault-tolerant machine may need many physical qubits to create a single high-quality logical qubit. The exact resource requirement depends heavily on the physical error rates, algorithm, error-correction code and desired reliability.
This means future quantum computers may contain many thousands, millions or more physical qubits to provide a much smaller number of extremely reliable logical qubits.
What Could Stop Quantum Computing From Scaling?
1. Physical errors
Even tiny error rates become significant when an algorithm contains huge numbers of quantum operations. Reducing physical errors is therefore essential.
2. Correlated errors
Error-correction systems work best when errors behave within the assumptions of the code. Rare correlated events can create more difficult failure modes. Nature’s report on Willow noted that rare correlated error events were still an important limitation in some experiments.
3. Cryogenic engineering
Superconducting qubits operate at extremely low temperatures. Scaling from laboratory systems to huge machines requires sophisticated cooling, wiring and control architectures.
4. Control electronics
Every qubit needs precise control and measurement. Increasing the number of qubits can dramatically increase the amount of electronics and data that must be handled.
5. Real-time decoding
Error correction is useful only if the system can detect and respond to errors quickly enough. Willow’s demonstrations included real-time decoding work, which is an important step toward practical fault tolerance.
6. Useful algorithms
Hardware alone is not enough. Researchers need algorithms that deliver meaningful results, and those algorithms must have a genuine advantage over the best classical alternatives.
Willow vs Sycamore: Why Google’s Progress Matters
Google’s Willow chip builds on the company’s earlier Sycamore quantum processor. In 2019, Google reported a quantum-computational demonstration that became one of the most famous moments in the field.
The key difference is that Willow is part of a broader push toward scalability and error correction. The field has gradually moved from asking whether quantum hardware can demonstrate a computational separation to asking whether that separation can become reliable, repeatable and useful.
Google demonstrated a famous beyond-classical benchmark using its earlier superconducting processor.
Google reported an important step toward logical qubits and error suppression.
The 105-qubit processor demonstrated below-threshold error correction and a dramatic RCS benchmark result.
Google announced a verifiable quantum-advantage experiment using Willow, reporting a 13,000× speed advantage for the selected comparison.
Why the 2025 Quantum Echoes Result Is More Interesting Than the 10²⁵-Year Headline
The 1025-year RCS comparison is spectacular, but it is intentionally a benchmark that becomes hard for classical computers. Quantum Echoes is more interesting from an application perspective because it attempts to connect quantum advantage with a structured scientific calculation.
Google described the experiment as verifiable quantum advantage. Verification matters because a scientific computation has to produce information that researchers can trust. If a quantum processor produces a result that cannot be independently checked or reproduced, its usefulness is limited.
Google’s 2025 announcement said Quantum Echoes could be useful for learning about the structure and dynamics of systems such as molecules and materials. The company also connected the work to a “molecular ruler” concept using nuclear magnetic resonance data.
These are still early-stage results, not a declaration that commercial quantum chemistry has arrived. But they show the direction the industry is trying to move toward: from quantum supremacy-style benchmarks to verifiable, scientifically meaningful computations.
What Willow Means for the Future of Computing
The most important lesson from Willow is not that classical computing has lost. It is that the definition of computing power may be expanding.
For decades, progress in computing was closely associated with making classical transistors smaller, faster and more energy efficient. GPUs then transformed massively parallel workloads and AI. Quantum computing introduces another computational model based on the laws of quantum mechanics.
If researchers can scale quantum error correction far enough, a quantum processor could become a specialized machine for problems that are fundamentally difficult for classical computers.
That could change scientific discovery itself. Instead of relying entirely on experiments followed by classical approximation, scientists could eventually simulate complex quantum systems directly and explore huge spaces of chemical or material possibilities.
The Most Important Number May Not Be 105
When people hear “Google Willow has 105 qubits,” the natural reaction is to compare that number with processor cores, GPU cores or transistor counts.
That is the wrong comparison.
The more important question is how many high-quality logical qubits a future system can produce from physical qubits, and how long those logical qubits can maintain useful information while a large algorithm runs.
In that sense, Willow’s most important contribution is not simply adding more physical qubits. It is showing evidence that error correction can improve as the system scales within the demonstrated regime.
Final Verdict: Did Google Willow Make Supercomputers Obsolete?
No — but it may have made the future of computing look very different.
Willow does not replace the world’s classical supercomputers. It does not run Windows, render movies, train today’s AI models in the same way as GPUs, or perform every scientific workload faster.
What Willow does demonstrate is far more subtle and arguably more important: a quantum processor can reach a regime where classical simulation becomes extraordinarily difficult, while simultaneously showing meaningful progress in quantum error correction.
The 2024 result established a major error-correction milestone. The later 2025 Quantum Echoes work pushed the story toward verifiable quantum advantage and scientific applications. Together, these developments suggest that quantum computing is slowly moving from a purely theoretical promise toward a technology that could eventually solve specific problems beyond the practical reach of classical machines.
The supercomputer era is therefore not ending. Instead, computing may be entering a hybrid era in which CPUs, GPUs, specialized accelerators and quantum processors each handle the problems they are best suited to solve.
Frequently Asked Questions
Is Google Willow a 105-bit processor?
No. Willow contains 105 physical quantum bits, or qubits. Qubits operate according to quantum mechanics and are fundamentally different from classical bits.
Can Willow replace my CPU or GPU?
No. Quantum processors are specialized systems. CPUs and GPUs remain essential for general-purpose computing and most current AI and scientific workloads.
What did Willow actually calculate in five minutes?
Google used a Random Circuit Sampling benchmark. It is a difficult quantum benchmark designed to test the ability of classical computers to simulate a quantum processor. It is not a normal commercial application.
What is the 10 septillion years figure?
Google estimated that the selected RCS computation would take about 1025 years on a leading classical supercomputer under the assumptions used for the comparison.
Why is quantum error correction so important?
Physical qubits are noisy and fragile. Error correction allows many physical qubits to work together as a more reliable logical qubit. Scaling quantum computers without improving logical reliability would not be enough.
What is a logical qubit?
A logical qubit is quantum information encoded across multiple physical qubits using an error-correction scheme. The goal is to make the logical qubit much more reliable than any individual physical qubit.
Does Willow already have practical commercial applications?
Willow represents major progress, but practical, commercially valuable quantum advantage remains a difficult goal. Google’s later Quantum Echoes work is an important step toward that direction, not proof that every commercial quantum application is ready today.
What could quantum computers eventually be used for?
Potential areas include quantum chemistry, materials science, drug discovery, optimization, simulation of physical systems, cryptography-related applications and selected machine-learning or sampling tasks.
Sources & Image Credits
Primary technical sources: Google Quantum AI’s official Willow announcement and the peer-reviewed Nature paper on quantum error correction below threshold. For the later Quantum Echoes development, Google’s October 2025 research announcement was also consulted.
- Google: Meet Willow, our state-of-the-art quantum chip
- Nature: Quantum error correction below the surface code threshold
- Google: Quantum Echoes and verifiable quantum advantage
- Google Quantum AI: Quantum Computer / Willow
Image licensing: The temporary microchip images used in this article are from Wikimedia Commons files released under CC0/Public Domain. They may be replaced later with your preferred images. CC0/Public Domain images do not require attribution, although keeping the source information is good practice.