Quantum Computers Have a Wiring Problem. This Tiny Frozen Chip Could Help

Quantum computers are often described as a race to build better qubits. Less attention goes to a stubborn engineering problem around them: every qubit needs precisely timed control signals, while the qubits themselves may sit inside a refrigerator colder than outer space. As machines grow, the cables running between room-temperature electronics and the frozen processor become a serious barrier.

A new cryogenic digital-to-analogue converter from Delft University of Technology points to another approach. Put more of the control electronics beside the qubits, and make those electronics work in the cold.

A control chip at 400 millikelvin

The prototype uses a 16 nanometre FinFET process and occupies only 0.026 square millimetres. Nature Electronics reported on July 20, 2026 that it consumed 275 microwatts and operated at 400 millikelvin, or about 0.4 degrees above absolute zero. It generated complex waveforms for 50 nanosecond two-qubit operations in a spin-qubit system, with a reported waveform fidelity of 99.99%.

A digital-to-analogue converter, or DAC, turns numerical instructions into the carefully shaped electrical pulses used to manipulate qubits. In ordinary computers this is routine electronic plumbing. Near a quantum processor it becomes unusually difficult. Transistors change behaviour at extreme cold, power creates unwanted heat, and tiny errors in pulse shape can reduce the quality of a quantum operation.

Why the cables become a bottleneck

Today's experimental machines can route many signals from warmer stages of a dilution refrigerator down to the processor. That strategy becomes harder as qubit counts rise. More wires occupy physical space, carry heat toward the coldest stage and complicate assembly. A refrigerator also has a limited cooling budget.

Cryogenic CMOS tries to move signal generation closer to the processor. Shorter connections could reduce latency and wiring, while familiar semiconductor manufacturing may make the control layer easier to integrate. Spin qubits are especially interesting because they can be fabricated in silicon structures that resemble technologies already used by the chip industry.

A promising part, not a solved computer

The result should not be confused with a demonstration of a large fault-tolerant quantum computer. The 99.99% figure describes the reported quality of the generated control waveform for a particular fast gate scenario. It does not by itself establish the end-to-end fidelity of a quantum algorithm, including the qubits, readout and environmental noise.

Power is also relative. A consumption of 275 microwatts is tiny on a desk, but cooling capacity is scarce at sub-kelvin temperatures. Replicating the circuit across thousands or millions of control channels could create a substantial heat load. Engineers must also manage interference, calibration, manufacturing variation and the connection between the control chip and the qubit array.

The bigger significance

Quantum progress depends on an entire stack: materials, qubits, control, error correction, packaging, cooling and software. A breakthrough in only one layer is not enough. This converter is interesting because it tackles the unglamorous infrastructure that could otherwise prevent promising qubits from scaling.

The likely future quantum computer will not be a naked quantum chip connected to a forest of laboratory instruments. It will need compact classical electronics operating at several temperature levels, each doing part of the work. This tiny frozen DAC is an early example of that architecture, and a reminder that the path to useful quantum computing may be decided as much by wires and heat as by qubits.