On November 12, 2026, Delft University of Technology and the quantum research institute QuTech announced that UniQ, a six-year research program aimed at scaling up semiconductor spin qubits, has been selected for a European Research Council (ERC) Synergy Grant. Four researchers specializing in materials science, cryogenic circuit engineering, and qubit control will work together on the project. In current experimental setups, multiple wires per qubit must be run in from outside the refrigerator, so as the number of devices grows, wiring space and heat generation become serious constraints. UniQ will test whether co-designing qubits and their peripheral circuits on a single chip can overcome these barriers to scaling.
Shrinking qubits alone won't produce logical qubits
Research on spin qubits has continued for years, driven by the expectation that the microfabrication techniques developed in semiconductor manufacturing can also be used for quantum computers. This approach, which uses the spin of electrons or holes to carry quantum information, may allow denser integration than methods such as superconducting qubits. However, even if current semiconductor manufacturing technology can form many quantum dots on silicon or germanium substrates, simply lining up physical qubits does not yield the "logical qubits" needed for stable computation. Protecting quantum information, which is vulnerable to noise from the environment, requires making many physical qubits interact appropriately, supplying precise control signals, and detecting errors without destroying the encoded information.
What has been announced is not a finished quantum processor but a joint research program that integrates materials science, cryogenic circuit engineering, and device control to explore the scaling of spin qubits. Project leader Menno Veldhorst explains that the control precision of spin qubits has steadily improved, and that understanding of how to connect and move qubits has also advanced. With the basic technology for operating individual qubits maturing, the idea is to move on to testing whole-system designs aimed at building logical qubits that support error correction.
In 2021, the Delft team demonstrated four germanium hole-spin qubits arranged in a 2×2 array in a paper published in Nature, "A four-qubit germanium quantum processor." In addition to programmable operations on one to four qubits, the experiment succeeded in generating a four-qubit GHZ state. However, this was a small-scale demonstration in a carefully tuned laboratory environment and does not mean the number of qubits can simply be increased. UniQ aims not to extend such foundational results in a straightforward way, but to reduce, through the design of the chip itself, the burden on the control system that grows as the number of devices increases.
Balancing heat and wiring density inside the refrigerator
One of the main factors hindering the scaling of spin qubits is the large number of wires linking the inside and outside of the dilution refrigerator. In typical experimental setups, a quantum chip cooled to cryogenic temperatures near absolute zero (roughly 100 mK or below) is connected to control and measurement equipment at room temperature by many cables. As qubits increase from dozens to hundreds, not only the space occupied by wiring but also the heat flowing in through the cables becomes impossible to ignore. Given the limited cooling capacity of the refrigerator, it is difficult to apply conventional wiring methods as they are to a large-scale quantum processor.
To address this problem, UniQ will place CMOS integrated circuits that operate at cryogenic temperatures (cryo-CMOS) close to the qubits and design the control system as an integrated whole. Specifically, the concept is to build circuits that distribute control signals and preamplifiers that boost weak readout signals onto the quantum chip, reducing the number of wires needed for external connections. However, placing electronic circuits near qubits turns those circuits into heat sources, which could undermine the stability of quantum states. In the millikelvin regime, the available cooling power is extremely small, and even slight heat from a circuit could affect qubit coherence.
Fabio Sebastiano, who is responsible for cryogenic electronics, explains that it is important to design qubits and electronic circuits so that each is easier for the other to work with. If the conditions for the signals needed to drive qubits can be relaxed, control circuits can be made smaller and consume less power. If low-heat circuits can be placed near the qubits, more qubits could be controlled without adding external wiring.
Supporting this design from the materials side is Giordano Scappucci's optimization of germanium structures. The aim is to create a low-noise, uniform crystalline environment in which qubits operate stably, while also forming the transistors needed for control circuits on the same material platform. By reducing impurities and defects in the material and suppressing unwanted electrical and magnetic noise, the goal is to reduce variation in characteristics from qubit to qubit. If device characteristics become uniform, the fine-tuning needed for each qubit may decrease, potentially lightening the burden on control circuits. However, how far improved material uniformity will simplify the overall system will need to be confirmed in future integration experiments.
Prior research has already shown how difficult it is to balance heat generation against control performance. A 2025 paper in Nature, "Spin-qubit control with a milli-kelvin CMOS chip," heterogeneously integrated a cryo-CMOS chip with about 100,000 transistors and silicon MOS spin qubits, demonstrating general-purpose quantum gate control with cryogenic circuits. When generating 100 mV pulses, the fixed power consumption of the digital circuitry was tens of microwatts, and the power consumption of the analog control cells was about 20 nW per MHz of operating frequency. Considering that the cooling capacity of a typical commercial dilution refrigerator at 100 mK is about 1 mW, the power that can be allocated to control circuits is limited.
In that study, control of single and two qubits using cryo-CMOS showed no major performance degradation compared with room-temperature control electronics. On the other hand, when the CMOS circuits were run at higher clock frequencies, a slight effect on qubit fidelity was observed, and it was suggested that heat flowing in from the circuits may have raised the electron temperature in the quantum device. In addition, because the tuning range of the tunnel coupling needed for state preparation and measurement was limited, the fidelity of two-qubit gates could not be quantitatively evaluated. Although integration of control circuits itself has been demonstrated, maintaining high control precision while managing heat appropriately remains an important challenge.
Furthermore, a 2021 paper in Nature Electronics, "A cryogenic CMOS chip for generating control signals for multiple qubits," measured six cryo-CMOS control cells operating at 100 mK and recorded an average power consumption of 18 nW per cell when generating 100 mV pulses. Based on this result, the research team estimated that 1,000 control cells could be cooled in a commercial dilution refrigerator. However, this is a scalability estimate calculated from power consumption; they did not actually control 1,000 qubits.
The positioning of these earlier studies and UniQ can be summarized as follows.
| 研究と実証状況 | 材料と集積手法 | 規模と測定条件 | 実証された成果 | UniQに関連する課題 |
|---|---|---|---|---|
| Delft 2021（実証済み） | ゲルマニウム正孔スピン量子ビット | 2×2配列、4量子ビット、ミリケルビン極低温 | 1〜4量子ビットの演算、GHZ状態生成 | 外部配線に依存する制御方式の大規模化 |
| Nature 2025（実証済み） | シリコンMOSスピン量子ビットとクライオCMOSの異種集積 | 約10万トランジスタのCMOS、100 mK級 | 汎用量子ゲート制御、アナログセル約20 nW/MHz | 高負荷時の発熱管理、2量子ビット忠実度の定量評価 |
| Nature Electronics 2021（実証済み） | クライオCMOS制御回路 | 6制御セルを100 mKで測定 | 100 mVパルス時にセル平均18 nW | 多数の制御セルと量子ビットの統合検証 |
| UniQ 2026（研究計画） | ゲルマニウム量子ビットと制御回路の一体設計 | 大規模アレイ、前段増幅器、信号分配回路の集積を構想 | 6年間の共同研究計画が採択 | 発熱抑制、素子均一性、量子ビット移動、誤り訂正への応用 |
Moving qubits to free up space for wiring and circuits
Integrating control circuits onto the chip requires space around the qubits for wiring and transistors. However, when spin qubits interact directly with one another, they usually must be brought within a few tens of nanometers, so the placement of control circuits and the connections between qubits compete for space. To address this, UniQ will use "shuttling," a technique that moves qubits between quantum dots. Rather than only letting qubits interact at fixed positions, moving them as needed is intended to give more freedom in chip layout.
Veldhorst's team will explore ways to improve how qubits are manipulated with voltage pulses and to relax the strict timing precision required of control circuits. If the demands on pulse timing become looser, the cryogenic circuits being developed by Sebastiano could be made smaller and consume less power. This would make it possible to fit more control functions into a limited chip area.
If qubits can be moved within an array, it becomes easier to connect devices that are far apart while also securing room between them for wiring and amplifier circuits. Moreover, higher connectivity between qubits could make it possible to adopt error-correction schemes that reduce the number of physical qubits needed to form a single logical qubit. However, the UniQ announcement does not give quantitative targets, such as which error-correcting code will be used or how much the number of physical qubits could be reduced.
Promising experimental results on qubit shuttling have already been reported. A 2024 paper in Nature Communications, "Coherent spin qubit shuttling through germanium quantum dots," demonstrated coherent transfer of spin qubits between germanium quantum dots. The effective transport distance was 312 μm for the spin ground state and 9 μm for coherent superposition states, the latter extending to 49 μm using echo pulses. However, these are effective distances derived from results of repeatedly moving qubits among a small number of quantum dots, and do not mean that a physical transmission path of that length was built. The research team also takes a cautious view of simply applying results obtained with a few quantum dots to long-distance connections, which would require greater uniformity and precise coupling control.
In addition, a 2025 paper in Nature Nanotechnology, "High-fidelity single-spin shuttling in silicon," used a conveyor-mode approach, in which an electron is confined in a moving potential minimum on a silicon substrate and carried along. It achieved a cumulative effective transport distance of 10 μm in under 200 nanoseconds, with an average spin-state fidelity of 99.5%. Spin coherence was also confirmed to improve compared with the conventional method of passing electrons sequentially between quantum dots. However, this is a result in silicon, and the same performance is not guaranteed in germanium, the platform at the center of UniQ.
The team led by Lieven Vandersypen will test whether these technologies can be combined on a single chip. UniQ will begin integration experiments at an early stage, measuring the noise introduced by electronic circuits placed near qubits and the changes in control signals passing through those circuits. The results will be fed back into Scappucci's materials development and Veldhorst's qubit design, with improvement and re-evaluation repeated. Vandersypen explains that improving the materials can simplify qubit control, and as a result the control circuits can also be made smaller. Rather than optimizing materials, devices, and electronic circuits separately, improving them with one another's constraints in mind will be the central research method of UniQ.
What can six years of research demonstrate?
UniQ's ultimate goal is to build a quantum processor chip in which preamplifiers and signal distribution circuits are integrated into an array of many spin qubits. Using this chip, the team will examine the connectivity structure of the qubits and the potential for applying quantum error correction. The program also envisions connecting multiple completed integrated chips as basic units to expand into larger quantum processors.
However, the announcement did not disclose specific performance targets. No quantitative metrics for judging processor performance were given, such as the number of qubits to be integrated, the chip's operating temperature, the allowable total power consumption, target quantum gate fidelity, or logical error rate. For that reason, future results will need to be evaluated by distinguishing whether the number of qubits merely increased, whether external wiring was reduced, whether high fidelity was maintained while the control circuits were operating, or whether the effectiveness of error correction was demonstrated.
According to the ERC announcement, the 2026 Synergy Grants drew 957 applications across all research fields, and 48 teams were selected. The success rate was about 5%, or roughly one in 20. A total of €500 million was allocated across the selected projects, for an average grant of about €10.4 million per project. However, this average is not necessarily the amount awarded to UniQ and should be distinguished from the individual confirmed grant.
As also summarized in a 2026 review in Nature Reviews Electrical Engineering, "Spin qubits in semiconductor platforms," many challenges remain in developing spin-qubit experimental techniques into standard mass-production semiconductor CMOS technology. Stable operation in the millikelvin regime, specialized material stack structures, variation in device characteristics, and integration with ultra-low-power control circuits are closely interrelated problems that must be solved.
What distinguishes UniQ is that it does not try to solve these challenges separately, but designs materials, circuits, and qubit-shuttling technology as a single system. The question is whether the project can go beyond raising qubit integration density and develop structures in which the qubits can actually be controlled and used for error correction. The results of the next six years of research will be an important input for judging the technical possibility that semiconductor spin qubits can move beyond small-scale laboratory demonstrations toward large-scale quantum processors.