Five silicon qubits ran together at about 99.9 percent accuracy on one control line

A team centered at Japan’s RIKEN Center for Emergent Matter Science has run five silicon spin qubits at the same time through one shared microwave control line, holding single-qubit gate accuracy at about 99.9 percent with all five active. With up to three qubits driven together, the same device stayed above 99.99 percent. The work appeared in the journal Nature Physics.

Tech news in one quick read: Get the most interesting stories in startups, tech and programming, each with a short TLDR, in one daily email from TLDR.

Subscribe to TLDR →

We may earn a commission if you sign up.

The figures describe one experimental chip and one kind of operation, single-qubit gates. They are not the accuracy of a full quantum computation, and they do not yet include the two-qubit gates a useful processor needs. What the result does show is a way around a problem that grows with every qubit added to a chip: each one normally wants its own control wiring.

Five qubits on a single microwave line

The paper, “Simultaneous high-fidelity single-qubit gates in a spin qubit array”, lists first author Yi-Hsien Wu and co-first author Leon Camenzind with colleagues at RIKEN, ETH Zurich, TU Delft and National Taiwan University, and with Seigo Tarucha among the authors. The preprint on arXiv describes the device as five spin qubits in an isotopically purified silicon-28 and silicon-germanium quantum well, all driven through one microwave line.

One line matters because wiring is a bottleneck. Quantum Zeitgeist’s account of the work says a shared line reduces the need for a dedicated control line per qubit, which simplifies the infrastructure and the cost. ScienceDaily’s summary calls shared microwave lines a practical route to larger processors. The catch is crosstalk, because driving one qubit through a common line disturbs its neighbors, and the new result is about making simultaneous gates work despite that, not merely one qubit at a time.

Camenzind, a RIKEN research scientist who supervised the experiments, said in the RIKEN release as reposted by ScienceDaily on Oct. 25, 2026: “This shows that high-fidelity control can be maintained even when several qubits are operated at the same time.” The repost trails the original announcement. Quantum Zeitgeist covered the work on Oct. 16, 2026, and Quantum Computing Report on Oct. 17, so the underlying release is older than the ScienceDaily date suggests.

The AC Stark shift and a pairwise fix

The authors trace the main error in parallel operation to the AC Stark shift. When a microwave pulse is applied to one qubit, it nudges the frequency of the others, and that accumulates as a phase error on qubits that were supposed to sit idle. The arXiv preprint separates two effects: z-phase errors between qubits from the Stark shift, and x-rotation errors from changes in the combined drive amplitude. Both are corrected by calibration.

The calibration cost is the notable part. Crosstalk phases add up linearly, so the correction can be built from pairs of qubits. The number of calibrations then scales as N times N minus one, instead of growing exponentially with every possible combination of qubits driven at once. First author Wu described the advantage as avoiding “an entirely separate calibration procedure for every possible combination of qubits.”

The pulse shape helped as well. According to the preprint, smooth Kaiser-window pulses pushed single-qubit gates above 99.99 percent on all five qubits, with some near 99.999 percent, while rectangular pulses reached only about 99.98 percent, limited by their abrupt edges. Quantum Computing Report’s account of the paper puts individual gate fidelities between 99.993 and 99.998 percent, and notes that earlier attempts at parallel driving had fallen to about 99.5 percent.

Drive-induced decoherence sets the ceiling

Quantum Computing Report lists the full five-qubit result as at least 99.90 percent, with one qubit reaching 99.977 percent, and compares it with the 99.9 percent level often cited as workable for surface-code error correction. The preprint puts it as “remain at the practical fault-tolerant threshold of 99.9% even during fully parallel five-qubit operation.” Four of the five qubits exceeded 99.9 percent at gate times of 250 and 500 nanoseconds.

The remaining shortfall comes from drive-induced decoherence. Fast gates need stronger microwave power, and stronger power degrades the qubits, probably through heating or microwave-induced noise, while slower gates suffer more from low-frequency dephasing. Camenzind called it a limitation but said “further improvements to device design could help mitigate this source of error.” The preprint suggests micromagnet changes that roughly double the driving-field gradient could allow faster gates at lower power.

The authors suggest the method could extend to about ten qubits. That is a modest step in count, from five to ten, but each added qubit on a shared line makes the crosstalk calibration harder, so it is a meaningful test. Quantum Zeitgeist reports that Tarucha, who manages the silicon quantum computing project, described the group’s long-term aim as a fault-tolerant quantum computer built on silicon spin qubits. Whether pairwise calibration and one shared line hold up beyond five qubits, and alongside two-qubit gates, is what the next device has to show.

The 99.9 percent figure should be read in that light. Roughly 99.9 percent with five qubits running at once is a result about control, measured on gates applied to single qubits. The above-99.99-percent figure with up to three qubits running shows how much margin exists when fewer qubits compete for the same microwave power, and the gap between the two is the drive-induced decoherence the team now wants to engineer away.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

Tech news in one quick read: Subscribe to TLDR →

We may earn a commission if you sign up.