A team from the University of New South Wales and imec reported single-qubit gate fidelity up to 99.8% for hole-spin qubits built on natural silicon. The peer-reviewed paper appeared in Nature Communications on August 27, 2026.
The result targets one of silicon quantum computing's strongest arguments: use semiconductor manufacturing methods instead of inventing a new production chain for every processor generation.
What the team measured
The device uses gate-defined quantum dots. Each dot confines holes inside a silicon structure, and electrical signals control the spin state. The researchers used fast control, exchange pulsing, and electric-dipole spin resonance to operate the qubits.
Single-qubit gate fidelity reached 99.8%. The paper also reports a two-qubit gate quality factor of 240. The authors interpret this figure as a physical-fidelity limit of 99.7% for the two-qubit operation.
Those numbers describe physical operations. They do not describe logical qubits, a fault-tolerant processor, or a useful algorithm running at scale.
Why natural silicon matters
Many silicon spin-qubit experiments use isotopically purified silicon to reduce noise from the spinful silicon-29 isotope. This work used natural silicon instead. Natural silicon contains silicon-29, so reaching high fidelity under those conditions removes one fabrication and supply-chain burden.
The device also follows a foundry-oriented path. The authors describe a platform with industrial-grade fabrication and a device structure close to one used in recent high-fidelity electron-spin work. A shared process flow matters because large quantum processors need repeatable devices, not one exceptional dot selected from a small batch.
Hole spins offer another engineering difference from electron spins. Their all-electrical control fits closely with conventional transistor-style wiring. Their p-type wavefunction also reduces sensitivity to nuclear-spin noise, according to the researchers' analysis.
The Diraq connection
Several authors hold affiliations with Diraq, the silicon spin-qubit company covered in our Chicago laboratory post and Santa Monica engineering hub post. The paper lists Andrew Dzurak as a Diraq director and CEO, and several authors disclose equity interests in Diraq.
That disclosure does not cancel the measurement. It gives readers useful context for separating the peer-reviewed device result from commercial expectations about future processors.
What the result does not show
A 99.8% single-qubit number is not a complete scalability result. A processor also needs high-fidelity two-qubit gates, reliable readout, uniform performance across many devices, fast control electronics, and an error-correction scheme with acceptable overhead.
The paper reports a two-qubit quality factor, not a large two-qubit algorithm or a logical-qubit demonstration. The device also does not show a large array operating with the same performance across every site. Those are the next tests for a silicon spin approach.
The natural-silicon result still matters. It puts hole-spin control close to the performance range needed for error-correction experiments while keeping fabrication tied to an established semiconductor ecosystem. The next milestone is not another isolated fidelity record. It is repeatable two-qubit operation across a larger array.