China’s new quantum computer hits stability milestone, beating Google on efficiency
Chinese team is first outside the US to cross key threshold that determines whether practical quantum computers can work reliably at scale
Chinese researchers have taken a major step in the global race to build practical quantum computers, becoming the first team outside the United States – and the second in the world after Google – to cross a key threshold that determines whether these machines can work reliably at scale.
A team led by Pan Jianwei at the University of Science and Technology of China said their superconducting quantum computer, Zuchongzhi 3.2, had reached the fault-tolerant threshold – a point where fixing errors made the system more stable rather than less, overcoming a long-standing problem in which the very process of error correction introduces new mistakes.
Their research, published last week in the journal Physical Review Letters, relied on microwave-based control rather than the hardware-intensive error-suppression methods used by Google. The Chinese approach “could offer a more efficient route than Google’s” to building large, fault-tolerant quantum computers, the team said in a statement on Monday.
Joseph Emerson, a physicist at the University of Waterloo in Canada who was not involved in the research, said the study tackled one of quantum computing’s most difficult problems: qubits drifting out of their intended states and quietly spreading errors through the system.
Writing in the American Physical Society’s Physics magazine, Emerson described the experiment as “an impressive feat”, while cautioning that it remained far from the scale needed for practical, real-world applications.
Quantum computers work by harnessing the laws of quantum physics rather than the simple on-off logic used by ordinary computers. In theory, this allows them to tackle certain tasks – such as optimising complex systems or simulating molecules – in minutes that would take today’s machines thousands of years to complete.
In practice, however, quantum computers face a fundamental obstacle: instability. Their building blocks, known as qubits, are extremely sensitive to heat, noise and tiny disturbances from their surroundings, causing errors to appear constantly during normal operation.
To manage this, scientists developed quantum error correction, which spreads information across many qubits and repeatedly checks for problems. But this creates a paradox: each extra qubit and each additional check also introduces new sources of error.
For years, attempts to correct mistakes made systems less reliable, not more.
That is why researchers focused on a critical tipping point known as the error-correction threshold. Below this threshold, error correction backfires and creates more mistakes than it removes. Above it, the balance flips and error correction delivers a net benefit, allowing systems to become more stable as they grow.
China and the US both began investing early in surface code quantum error correction, one of the most widely studied methods for protecting quantum information. In 2022, Pan’s team used an earlier processor, Zuchongzhi 2, to achieve a minimal error-correcting unit, known as a distance-3 surface code logical qubit, as an initial proof of principle.
The following year, Google pushed the technique further by achieving distance-5 surface code error correction. But in both cases, relatively high error rates in the underlying qubits prevented the systems from truly crossing the threshold.
That changed in February, when Google reported a breakthrough using its Willow quantum processor. By suppressing a particularly harmful class of errors known as leakage using direct current pulses, Google became the world’s first team to achieve a distance-7 surface code logical qubit operating below the threshold.
This approach, however, places tight constraints on chip design and requires increasingly complex wiring in ultra-low-temperature environments as systems scale up, according to Pan’s team.
In the new study, the Chinese researchers took a different path. Working with the 107-qubit Zuchongzhi 3.2 processor, they developed an all-microwave method to suppress leakage errors, using carefully timed microwave signals instead of additional hardware controls.
Combining this approach with surface code error correction, Pan and coworkers were able to build a distance-7 logical qubit, matching the scale of Google’s most advanced demonstrations. They found that as the system grew larger, the overall error rate fell rather than rose.
The researchers measured an error-suppression factor of 1.4, meaning each increase in the size of the error-correction code reduced errors instead of amplifying them, which was clear evidence that the system was operating below the threshold.
They said that the all-microwave approach could offer practical advantages as quantum computers grow larger. Because microwave signals can be multiplexed, allowing multiple signals to travel along a single wire, the method might reduce wiring complexity and hardware overhead, two major obstacles to scaling quantum processors.
Taken together, they said, the results pointed to a more flexible and potentially more scalable route towards fault-tolerant quantum computers with hundreds of thousands or even millions of qubits.