Reducing Wires in Quantum Computers
As quantum computers continue to grow in size, one of the bottlenecks is the number of control wires that need to be connected to the quantum bits (qubits). A new theoretical study explores so-called time multiplexing, where one wire controls several qubits [1]. The researchers found that although this strategy requires extra processing time, the delays are less than expected, in part because control signals can be scheduled when certain qubits are busy with computations. The results could spur development of the electronic switches needed for time multiplexing in superconducting quantum computers.
Many state-of-the-art quantum computers consist of 100 or more superconducting qubits that operate inside dilution refrigerators at temperatures near absolute zero. Photos of these devices often show a tall, shiny column filled with dozens and dozens of connected wires—which might be mistaken for the qubits. Instead, these wires carry microwave signals from the room-temperature electronics that control the quantum processors to the micrometer-sized qubits inside the cryogenic refrigerator. The number of control wires can limit increases in the sizes of quantum computers. “You would like to have one wire going down to each qubit,” says Anton Frisk Kockum from Chalmers University of Technology in Sweden. “But that takes up a lot of space and brings heat into the fridge.”
One way to get around this constraint is to use multiplexing, in which multiple qubits are connected to a single control wire. Two types of multiplexing exist. Frequency multiplexing involves assigning a particular microwave frequency to each qubit. The problem with this approach is that it’s difficult to tune qubits to prevent responses to signals intended for their neighbors. Time multiplexing involves sending control signals at different times into a single wire that splits into multiple wires inside the fridge. A router, or switch, directs each signal to its target qubit.
This second strategy was thought to slow down a quantum computer. “The signals are normally sent simultaneously, but with time multiplexing they would be sent one after the other,” says Chalmers team member Marvin Richter. If, for example, one wire connects to four qubits, then a computation would presumably take 4 times longer. Such a delay would be detrimental to performance, Kockum explains, as noise tends to build up in qubits over time.
Researchers have been developing multiplexing technology in labs, but the impact on performance has been unclear. Kockum, Richter, and their colleagues have now investigated the cost of time multiplexing in terms of computational slowdown. They imagined a generic quantum computer setup with a switch at the end of each control wire. In the setup, microwave signals arrive one after another in a series, and the switch directs each signal to the appropriate qubit.
The researchers considered the typical architectures of current superconducting quantum computers, which include a combination of two types of logic gates (the basic elements of a processor): single qubit and two qubit. Two-qubit gates naturally take longer to operate. The researchers found that they could schedule faster single-qubit control signals to other qubits on a shared wire while the slow two-qubit gates completed their operation. “We can hide the effect of multiplexing behind the two-qubit gates,” Kockum says.
The analysis showed that the cost in processor speed depends on the way that the qubits are connected. But in general, the researchers found that the slowdown was less than expected, and in some cases it would be negligible. “You can schedule these signals in a nice way such that qubits perform [only] a little slower than if each had its own wire,” Kockum says. The researchers believe that these results will motivate other quantum computing research teams to develop the switches that are needed to perform time multiplexing.
“It’s quite nice work, addressing the central scalability challenge in cryoenabled quantum computing,” says microelectronics expert Mika Prunnila from VTT Technical Research Centre of Finland. The expectation was that time multiplexing would incur “large execution-time penalties,” so it is surprising that the cost is relatively modest, Prunnila says. Ming Yuan, a quantum computing researcher from the University of Colorado Boulder, says that Kockum and colleagues have done “a very comprehensive study” showing that it’s possible to efficiently arrange gate sequences for many quantum tasks. “I think this [multiplexing strategy] may serve as an alternative plan for the superconducting qubit platform,” Yuan says.
–Michael Schirber
Michael Schirber is a Corresponding Editor for Physics Magazine based in Lyon, France.
References
- M. Richter et al., “Overhead in quantum circuits with time-multiplexed qubit control,” PRX Quantum 7, 020308 (2026).