China's Zuchongzhi 3.0 Claims a Quadrillionfold Quantum Speed Advantage

105 qubits and a calculation estimated at 6.4 billion years

Chinese scientists at the University of Science and Technology of China (USTC) have unveiled Zuchongzhi 3.0, a quantum processor with 105 superconducting qubits that claims to outperform the world's fastest supercomputers by a factor of one quadrillion on a specific computing benchmark. The achievement places the Chinese team alongside Google at the front of an international contest to build more powerful quantum machines.

The central experiment used 83 of the processor's qubits to run a random circuit sampling task through 32 cycles. Zuchongzhi 3.0 completed that task in a few hundred seconds. The researchers' estimate puts the time needed to simulate the same task on the Frontier supercomputer at approximately 6.4 billion years.

A quadrillion means a million billion. The comparison concerns the particular sampling task used in the experiment: researchers run sequences of quantum operations and measure the patterns they produce. Practical uses in chemistry, logistics and other fields remain later goals on the team's roadmap.

As of August 28, 2025, the breakthrough has drawn attention from governments, industry leaders and researchers, while strengthening China's position in a field where US companies have set prominent benchmarks. Google's latest Willow processor also has 105 qubits, making the quality of those qubits and the experiments performed with them central to the comparison.

Zuchongzhi 3.0 combines a larger processor with changes to materials, manufacturing and control accuracy. Its achievement also brings an engineering problem into sharper focus: quantum information is fragile, and a machine must preserve it long enough to finish its calculations.

What is inside the new processor?

Zuchongzhi 3.0 contains 105 transmon qubits arranged in a rectangular grid measuring 15 by 7. A qubit is the basic unit of information in a quantum computer. Transmons are a type of superconducting qubit, and the team's processor uses tantalum, niobium and aluminum to help reduce noise and improve stability.

Noise means disturbances that interfere with the quantum information being processed. Reducing it helps qubits retain their quantum state, a property called coherence. Longer coherence gives the machine more time to carry out operations before that information deteriorates.

The processor has a reported coherence time of up to 72 microseconds and a dephasing time of 58 microseconds, figures described as world class. Dephasing concerns the loss of the timing relationship within a quantum state. Both measurements help describe how long the delicate information can remain usable.

Seventy two microseconds is a tiny interval in everyday life, yet quantum operations happen fast enough for improvements at this scale to matter. The team achieved those improvements through changes to qubit design, fabrication and the process used to connect the chip's components.

How accurately can the chip perform operations?

Zuchongzhi 3.0 reports a fidelity of 99.90% for operations involving one qubit and 99.62% for operations involving two qubits. Gate fidelity measures how accurately an intended quantum operation is performed. A gate is a basic step in a quantum calculation, comparable to an instruction that changes the information being processed.

The two measurements matter because a large calculation requires many operations, including interactions between qubits. Small errors can accumulate as a circuit grows longer. Higher accuracy gives researchers more room to run complex experiments before mistakes overwhelm the result.

The reported fidelities rival or surpass those of Google's latest quantum chips in some measures. Achieving operational fidelities above 99.9% is an important goal for practical quantum computing, although Zuchongzhi 3.0's stated fidelity for operations involving two qubits remains below that level.

Why random circuit sampling is the central test

Random circuit sampling, usually shortened to RCS, involves applying a complex sequence of quantum operations and measuring the resulting output distribution. In simple terms, researchers ask a quantum machine to generate results from a circuit, then examine how often different results appear.

Simulating that process on a classical computer becomes increasingly difficult as the number of qubits and the circuit depth increase. Circuit depth describes how many successive layers of operations the machine runs. Zuchongzhi 3.0's experiment combined 83 qubits with 32 cycles, creating a demanding task for classical simulation.

RCS is widely used as a benchmark for quantum computational advantage, also called quantum supremacy. The term describes the point at which a quantum machine can perform a task that is effectively beyond the reach of even the most powerful classical supercomputers.

The benchmark is designed to be difficult for classical machines and manageable for quantum ones. Its immediate purpose is to measure that advantage. Most everyday computing needs, and the practical industrial applications researchers hope to develop, involve different tasks.

Classical simulation methods continue to improve, raising the standard that quantum experiments must meet. The race therefore involves both stronger quantum hardware and better algorithms for ordinary computers. Critics question how much RCS performance will translate into useful applications, while researchers continue work on hardware, error correction and algorithms.

How Zuchongzhi compares with Sycamore and Willow

Google's Sycamore processor drew worldwide attention in 2019 when it performed a sampling task in 200 seconds. Google estimated that a classical computer would need 10,000 years to do the same work. Zuchongzhi 3.0's reported advance widens the computational gap by six orders of magnitude, or a factor of one million.

Google introduced Willow in late 2024. Like Zuchongzhi 3.0, it has 105 qubits. Google's claim for its RCS experiment puts the classical simulation time at 10 to the power of 25 years, a number vastly larger than the estimate attached to Zuchongzhi's experiment.

The experiments use different setups and circuit complexities, so their headline simulation times depend on the particular task each machine performed. Comparing the processors also requires examining coherence, gate accuracy and the scale of their sampling experiments.

Comparative assessments give Willow a slight advantage in some qubit quality measures, including coherence times and gate fidelities. Zuchongzhi 3.0 matches or exceeds it in others, including the scale and speed of its RCS experiments. Both processors are among the leading quantum hardware systems, pushing the limits of current technology.

The main milestones span several years:

• 2019: Google's Sycamore completed a sampling task in 200 seconds, with a claimed classical simulation time of 10,000 years.
• Late 2024: Google introduced its 105 qubit Willow chip, with an RCS claim equivalent to 10 to the power of 25 years of classical simulation.
• By August 28, 2025: USTC's Zuchongzhi 3.0 had demonstrated an 83 qubit experiment with 32 cycles in a few hundred seconds, with an estimated Frontier simulation time of approximately 6.4 billion years.

The engineering behind the larger chip

Zuchongzhi 3.0 increases the processor size from the 66 qubits of Zuchongzhi 2.0 to 105. That is 39 additional qubits, an increase of about 59%. Building a larger grid also requires preserving the accuracy and stability of its components as more qubits interact.

Tantalum, niobium and aluminum play a role in reducing noise and improving coherence. Those material choices support longer and more complex calculations by helping the processor retain quantum information.

The team also used an indium bump process to bond two sapphire chips in a flipped arrangement, each containing part of the qubit array. The process allows precise alignment and integration while reducing contamination and improving signal integrity. Reliable signals are necessary to control the qubits accurately.

Changes to capacitor pad design, Josephson junction parameters and wiring layouts further reduced errors and improved stability. These are parts of the electrical structure that shape how the superconducting qubits behave and how researchers control them.

Together, the manufacturing and design changes have produced operational fidelities and error rates described as setting new industry standards. They also provide a foundation for building quantum systems that are more reliable and can grow to larger sizes.

Zhu Xiaobo's roadmap has three stages

Zhu Xiaobo, chief designer of Zuchongzhi 3.0, describes a progression from beating supercomputers on particular tasks to developing machines capable of much broader work. His stated roadmap connects the current benchmark achievement with practical applications and, eventually, universal computing.

Our roadmap involves not only surpassing classical supercomputers but also finding practical quantum applications and ultimately achieving universal, fault-tolerant quantum computing.

The three stages are:

• Surpass classical supercomputers on specific tasks, establishing quantum supremacy.
• Find practical quantum applications in optimization, chemistry and machine learning.
• Achieve universal, fault tolerant quantum computing capable of solving a wide range of problems.

Fault tolerance means keeping a calculation reliable despite errors in the underlying hardware. Universal quantum computing means supporting a broad range of quantum calculations rather than concentrating on a particular benchmark.

Zuchongzhi 3.0 represents progress in the first stage. Estimates place universal, fault tolerant quantum computing at least 15 years away, while practical quantum applications may emerge within the next decade. These are expectations about different stages of development; no fixed date has been given for reaching them.

Errors and scale remain the next technical challenges

The Zuchongzhi team is researching surface code error correction, with the aim of increasing code distances and improving fault tolerance. Quantum error correction uses multiple physical qubits to protect quantum information against mistakes. A greater code distance means a stronger level of protection against combinations of errors.

Quantum systems are inherently fragile, and errors can accumulate during complex computations. Effective correction is therefore essential for moving from a successful sampling experiment to longer calculations that people and businesses can use.

Building processors with hundreds or thousands of qubits introduces another challenge: maintaining coherence and low error rates across the entire device. Materials, chip design and fabrication methods must continue improving as the systems grow.

Quantum machines also need to work smoothly with classical computing infrastructure. Ordinary computers will remain part of the surrounding system, so useful applications depend on reliable integration as well as quantum processing.

Algorithm development is another part of the team's work. Researchers need methods that solve practical problems faster than classical alternatives. The team is exploring quantum error correction, entanglement, simulation and quantum chemistry as it develops the foundations for later generations of quantum technology.

Why China and the United States are investing

China and the United States have both invested heavily in quantum research, recognizing possible economic, scientific and military uses. Zuchongzhi 3.0 strengthens the competition between Chinese research teams and US technology companies, particularly Google.

IBM is another major participant, pursuing a different approach within superconducting qubit technology. Chinese researchers are also developing photonic quantum computers, including the Jiuzhang series. Photonic systems use light for quantum processing, giving the field another route alongside superconducting devices.

Each approach has its own strengths and technical challenges. New breakthroughs appear regularly, making leadership depend on more than a single experiment or processor size.

The US Department of Defense and other agencies have described China's rapid progress in quantum technology as a potential strategic threat. Their concerns include work on quantum imaging, navigation, radar and secure communications. The unveiling of Zuchongzhi 3.0 is expected to encourage further investment and research worldwide.

Where useful quantum computing could make a difference

Cryptography is one of the areas attracting government attention. Future quantum computers could break widely used encryption schemes, creating a need for quantum resistant cryptography. That prospect connects quantum research with cybersecurity and national security planning.

Drug discovery and materials science are other potential uses. Quantum simulations could model complex molecules and chemical reactions, helping researchers develop medicines or design materials with desired properties for energy, electronics and manufacturing.

Optimization concerns finding a good solution among many possible choices. Quantum algorithms could eventually make some of these tasks more efficient in logistics, finance and supply chain management. Quantum machine learning could also introduce new approaches to data analysis and pattern recognition.

These possibilities remain largely theoretical. Zuchongzhi 3.0's benchmark success provides a hardware milestone, while useful applications depend on progress in reliability and the algorithms needed for particular problems.

Researchers see a foundation for further progress

Peer reviewers and experts have praised Zuchongzhi 3.0 as a substantial upgrade from earlier devices, citing its leading performance and engineering sophistication. The processor combines more qubits with efforts to control noise and improve accuracy, all necessary for more demanding experiments.

Nobel laureate Frank Wilczek described the wider potential of quantum tools to support further technological development:

These quantum tools will enable the development of more sophisticated quantum technologies, fueling further innovation and breakthroughs.

USTC's work brings hardware performance, materials and error correction into the same development effort. The immediate achievement is a powerful sampling experiment. The team's longer ambition is to carry that progress into useful applications and, eventually, reliable quantum computing across a wide range of tasks.

The Bottom Line

• Zuchongzhi 3.0 has 105 superconducting transmon qubits arranged in a grid measuring 15 by 7.
• Its 83 qubit experiment with 32 cycles took a few hundred seconds, with an estimated Frontier simulation time of approximately 6.4 billion years.
• The claimed speed advantage of one quadrillion concerns a random circuit sampling benchmark.
• The chip reports coherence of up to 72 microseconds, dephasing of 58 microseconds, and gate fidelities of 99.90% for one qubit and 99.62% for two qubits.
• Google's Willow also has 105 qubits, with comparisons depending on hardware quality and the particular experiment.
• The team is pursuing surface code error correction, practical applications and eventually universal, fault tolerant quantum computing.
• Potential uses include cryptography, drug discovery, materials science, optimization and artificial intelligence, while major engineering and algorithm challenges remain.