快猫短视频

The number of qubits that have been entangled in quantum computers has nearly doubled within the past year 鈥� the increase is happening so fast, it seems to be following a 鈥渜uantum Moore鈥檚 law鈥�.

First proposed by Gordon Moore at Intel in 1965, Moore鈥檚 law states that the power we can get out of a single traditional computer chip doubles at regular intervals; every year at first, then every two years as manufacturing encountered new difficulties. Now, we may be seeing the beginning of a quantum version of the trend.

To fulfil the promise of solving problems that are intractable for the best conventional supercomputers, quantum computers must include more qubits. But how those qubits are connected to each other also matters, especially when it comes to entanglement, a uniquely quantum connection that coordinates their behaviour.

鈥淓ntangling more qubits isn鈥檛 just about breaking records, it鈥檚 about expanding the frontier of what quantum computers can actually do. It鈥檚 a sign that quantum hardware is maturing and getting closer to practical applications,鈥� says at the聽University of Science and Technology of China.

The records are being smashed left and right. In December 2024, the entanglement record was held by 50 qubits. But recently, a team led by at the quantum software firm Q-CTRL combined 75 superconducting qubits into an entangled state called a Greenberger-Horne-Zeilinger (GHZ) state. Around the same time, Zhu and his colleagues entangled 95 qubits within the superconducting Zuchongzhi quantum computer into a 鈥渃luster state鈥�. Meanwhile, 聽at IBM and his colleagues .

at Q-CTRL says that entanglement is part of the 鈥渢he secret sauce of quantum computers鈥�, but it is also one of their most finicky properties. 鈥淚t鈥檚 very hard to create, but it鈥檚 even harder to maintain,鈥� he says. Creating highly entangled states plays a central role in running many quantum algorithms, says Baum.

Subscriber-only newsletter

Sign up to Lost in Space-Time

Sign up to newsletter

To make their entangled qubit state stable and reliable, the Q-CTRL researchers used novel methods for both detecting errors and suppressing them. All existing quantum computers are prone to errors, so this meant that some runs of the experiment weren鈥檛 successful 鈥� about 80 per cent of the time, the quantum computer detected an error that it had not managed to suppress.

Baum says this may sound severe, but it is an improvement on past studies and means that a calculation on the quantum computer would simply take more time because it may have to be repeated. Often, people combat this by adding extra qubits to correct errors 鈥� compared to that, this strategy keeps the resources needed to a 鈥渞ather reasonable鈥� level, says Liao.

IBM鈥檚 team also used an error detection approach: within a quantum computer with 156 qubits, 120 were entangled and the rest were used to flag when the process went awry. 鈥淲e use these extra qubits to detect errors and to protect the main computational qubits,鈥� says Javadi-Abhari. In this way, the team could get 鈥渕ore computation鈥� out of hardware that is not yet fully immune to errors, he says.

The extra qubits play an important role in making this useful for computing, but they also make it challenging to fairly compare the performance and usefulness of various quantum computers. at the University of Innsbruck in Austria says that though the numbers of qubits in these experiments are increasingly large, what matters more is how they are used. At the same time, the demonstration of control on these larger quantum systems is progress in itself, he says.

鈥淚 think this 鈥榪uantum Moore鈥檚 law鈥� is happening because the entire ecosystem around quantum computing is improving, not just the number of qubits but everything that supports them,鈥� says Zhu. As for whether it will continue, he is optimistic. In fact, using advanced software tools such as AI-driven optimisation may even speed it up, he says.