A team from RIKEN, National Taiwan University, ETH Zurich, and TU Delft has, according to Quantum Computing Report, simultaneously controlled five silicon spin qubits over a shared microwave line with high accuracy; the study appeared in Nature Physics. When controlling each qubit individually, the fidelities of primitive pi/2 gates were, according to the study abstract, well above 99.99%, in some cases approaching 99.999%. During simultaneous operation of up to three qubits, fidelities remained above 99.99%; when operating all five qubits in parallel, they reached 99.9%, which the authors describe in the study abstract as a practical threshold for error tolerance. The examined groups of three were operated with a pulse duration of 250 nanoseconds; the five-qubit values applied to 250 and 500 nanoseconds.
Why a Shared Line Is Difficult
The silicon spin qubits examined in the study store quantum information in the spin of an electron trapped in a quantum dot, as FreeQuantumComputing describes. They are considered a promising design for quantum computers because they can be manufactured using the same semiconductor manufacturing technologies as today's computer chips, as The Quantum Insider reports. Silicon spin qubits typically lose accuracy when controlled in parallel. Individual microwave lines for each qubit create problems with line routing and heat in growing processors; a shared line reduces hardware requirements, but simultaneous pulses cause crosstalk. According to the study abstract, the main cause of this accuracy loss in the examined five-qubit array is a so-called AC Stark shift, by which microwave signals change the effective energy levels of a qubit and shift the phase of its quantum state. According to the study abstract, the mitigation protocol is scalable; first author Yi-Hsien Wu explained that the method uses calibrations between qubit pairs instead of requiring a calibration for every possible combination.
In 2025, a team around Paul Steinacker achieved single and two-qubit control fidelities above 99% at four two-qubit devices manufactured industrially at imec and identified residual isotopes with nuclear spin as a significant error source. Unlike both studies examining a two-qubit processor and four two-qubit devices respectively, the new study shows comparable fidelities for up to five qubits over a shared line, namely for single-qubit gates.
What the Technology Cannot Yet Do
Spin-valley mixing disrupted two-qubit gate operation, according to FreeQuantumComputing. The remaining accuracy loss is attributed to drive-induced decoherence from higher microwave power in parallel operation; researcher Camenzind, who participated in the study, explains that improvements to the device could mitigate this effect.
What Is Missing for More Than Five Qubits
Project manager Seigo Tarucha, according to The Quantum Insider, names the construction of a fault-tolerant quantum computer from silicon spin qubits as a long-term goal, with reliable parallel control as an essential step. Camenzind sees pairwise calibration of a major error source as supporting shared control lines as a practical approach to scaling quantum processors. According to FreeQuantumComputing, the publication estimates that the architecture with gates of 500 nanoseconds could control approximately ten qubits with a fidelity near 99.9 percent; this estimate is not a demonstration with ten qubits. Until such a module actually controls more than five qubits and two-qubit gates with high fidelity are demonstrated, the path from five precisely controlled qubits to a fault-tolerant quantum processor remains open, according to FreeQuantumComputing.