A five-qubit silicon array controlled through one shared microwave line offers a practical step toward scaling quantum processors.
Quantum computers need more than qubits that work well one at a time. They must control many qubits together without signals meant for one disrupting its neighbors. A new five-qubit silicon experiment takes a step toward that goal: researchers used one shared microwave line to control the array, keeping accuracy above 99.99% when up to three qubits operated simultaneously.
The work, published October 15 in Nature Physics by researchers including the RIKEN Center for Emergent Matter Science, addresses a practical hurdle in building larger quantum machines. Giving every qubit its own control wire can become unwieldy as a system grows. A shared line offers a more compact approach, but it also raises the risk that control signals interfere with one another.
A qubit is the quantum counterpart of a computer bit. Conventional bits are either 0 or 1; qubits can occupy combinations of states. That flexibility is useful for quantum algorithms, but the states are delicate. Unwanted disturbances can introduce errors before a calculation finishes, so processors need both precise operations and ways to detect and manage faults.
The RIKEN team traced a key source of accuracy loss to AC Stark shifts. Microwave signals used to control a qubit can also shift its energy levels, changing how it responds. During simultaneous operations, those shifts can throw off carefully timed gates. The researchers mitigated the effect with pairwise phase calibrations, adjusting the relative phases of signals applied to qubit pairs.
Individual gate fidelities were well above 99.99%, with some approaching 99.999%, according to the institutional release. Fidelity remained above 99.99% when as many as three qubits operated at once. With all five operating simultaneously, it was approximately 99.9%. The decline shows why results from a single qubit cannot, by themselves, establish that a processor will work reliably at scale.
The researchers also identified a remaining limitation: stronger microwave drive can cause decoherence and heating, putting quantum states at risk. The authors said stronger magnetic-field gradients built into the device could deliver comparable gate speeds at lower power. That proposal still requires testing, and the team will need to show that high accuracy persists in larger arrays and more complex operations.
The silicon result is one of two recent efforts aimed at different quantum-computing bottlenecks. Google Quantum AI has reported a protocol called magic state cultivation on its Willow processor. Google’s described engineering advance included switching from a distance-3 color code to a distance-5 surface code, with cultivated states surviving the transfer. Magic states are needed for many general-purpose, fault-tolerant quantum algorithms, making their preparation an important part of the path to useful machines.
That Google result concerns state preparation, not a 40-fold improvement in overall processor performance. Google has said larger-scale demonstrations and further technical work are still needed. The distinction matters: preparing a valuable quantum state is not the same as running a large, reliable calculation.
Together, the efforts show how progress depends on solving separate technical problems: controlling qubits in parallel and preparing the resources algorithms require. Neither demonstrates a practical, large-scale quantum computer. The next measure of progress will be whether these methods can scale while preserving accuracy and limiting the noise and heating that undermine quantum states.

