Dutch Quantum Breakthrough: New Chip Tests Over 1,000 Qubits at Once
Delft, Friday, 13 February 2026.
QuTech researchers developed QARPET, a revolutionary chip architecture that can test 1,058 semiconductor spin qubits simultaneously in a single cooldown. The crossbar design achieves an unprecedented density of 2 million qubits per square millimeter while using only 53 control lines, marking a crucial advance toward scalable quantum computing manufacturing.
Quantum Computing’s Scalability Challenge Meets Semiconductor Solution
This breakthrough addresses quantum computing, specifically the development of semiconductor spin qubits - quantum bits that leverage the intrinsic angular momentum of electrons trapped in semiconductor materials [1]. Unlike traditional quantum computing approaches that rely on superconducting circuits or trapped ions, semiconductor spin qubits offer a critical advantage: compatibility with existing silicon manufacturing processes used in conventional electronics [GPT]. The QARPET (Qubit-Array Research Platform for Engineering and Testing) platform reported in Nature Electronics on February 12, 2026, represents a fundamental shift in how quantum processors can be tested and scaled [2]. The device was engineered specifically to tackle what Associate Professor Giordano Scappucci at TU Delft calls a key challenge: “how to efficiently evaluate large numbers of qubits, especially as devices with millions of quantum bits get developed” [1].
Engineering Marvel: From Concept to Reality
The fabrication of QARPET pushed nanomanufacturing to its limits, according to Alberto Tosato, the engineer who designed the first layouts [1]. “When I designed the first layouts, I honestly did not expect them to work,” Tosato explained, noting that “the number of crossing electrodes is extremely high. It pushes the limits of nanofabrication, we saw it as a test that would probably fail” [1]. The device structure, when viewed under a microscope, appears almost woven due to its complex crossbar architecture [1]. The successful demonstration at millikelvin temperatures - around 100 mK - proved that the ambitious design could function in the extreme conditions required for quantum computing operations [5]. The chip was fabricated using a germanium/silicon-germanium (Ge/SiGe) semiconductor structure, incorporating 23 by 23 tiles arranged in a crossbar layout [2].
Unprecedented Density and Scalability Achievements
The QARPET architecture achieves remarkable efficiency metrics that set new benchmarks for quantum device testing. The crossbar design enables a qubit density of 2.000 million qubits per square millimeter while requiring only 53 control lines to manage the entire 23×23 array [5]. This sublinear scaling of interconnects represents a crucial engineering breakthrough, as traditional approaches would require exponentially more control infrastructure [GPT]. Each tile in the array has a footprint of just 1 micrometer and contains two spin qubits plus one charge sensor, demonstrating the platform’s compact design [2]. The researchers successfully tested 40 tiles from the full array, achieving tile addressability in 38 out of 40 cases and demonstrating single-hole quantum dot functionality in 37 tiles [5]. Statistical analysis revealed that applying the median voltage settings would achieve single-hole occupation in 25% of cases, increasing to 75% when targeting up to five holes per quantum dot [5].
Performance Validation and Future Implications
The team’s rigorous testing methodology validated key performance metrics across the QARPET device. Charge noise measurements across functional tiles yielded a geometric mean of 2.4 ± 1.7 microelectron-volts per square root hertz, with the best-performing tile achieving noise levels as low as 0.36 microelectron-volts per square root hertz [5]. Coherence time measurements using both Ramsey and Hahn-echo experiments demonstrated T2* times of 4.43 ± 0.13 microseconds and 5.75 ± 0.19 microseconds for the two measured qubits, with echo-enhanced coherence times reaching 12.69 ± 0.40 microseconds [5]. These results position semiconductor spin qubits competitively within the broader quantum computing landscape. As quantum computing expert James Wootton noted on LinkedIn, this achievement represents a milestone that superconducting qubits and trapped ions surpassed a decade ago, marking spin qubits’ entry into a new era of scalability [4]. The research team, led by Scappucci at QuTech - a collaboration between TU Delft and TNO based in Delft, Netherlands - plans to integrate machine learning-assisted routines for autonomous device tuning and control [5].