It’s hard to mix electronic manufacturing and flexible geometry.

To get quantum computing to work, we will ultimately need lots of high-quality qubits, which we can tie together into groups of error-corrected logical qubits. Companies are taking distinct approaches to get there, but you can think of them as falling into two broad categories. Some companies are focused on hosting the qubits in electronics that we can manufacture, guaranteeing that we can get lots of devices. Others are using atoms or photons as qubits, which give more consistent behavior but require lots of complicated hardware to manage.

One advantage of systems that use atoms or ions is that we can move them around. This allows us to entangle any qubit with any other, which provides a great deal of flexibility for error correction. Systems based on electronic devices, in contrast, are locked into whatever configuration they’re wired into during manufacturing.

But this week, a new paper examined research that seems to provide the best of both worlds. It works with quantum dots, which can be manufactured in bulk and host a qubit as a single electron’s spin. The work showed that it’s possible to move these spin qubits from one quantum dot to another without losing quantum information. The ability to move them around could potentially enable the sort of any-to-any connectivity we see with atoms and ions.

Quantum trade-offs

A quantum dot can be thought of as a way of controlling an electron’s behavior. Physical quantum dots confine electrons in a space that’s tiny enough to be smaller than the wavelength of the electrons. Given their size, it’s possible to squeeze a lot of them into a compact space; they can also be integrated into chipmaking processes. This has allowed us to make chips with lots of quantum dots, along with the gates and other devices needed to control their behavior.

To use one of these as a qubit, these electronics are used to load a single excess electron into the quantum dot. Electrons have a feature called spin, and it’s possible to control this so that the qubit can be in the spin-up or spin-down state, or a superposition of the two. While qubits based on electrons tend to be relatively fragile—it’s pretty easy for the environment to knock electrons around a bit—the quantum dots tend to keep them isolated from the environment enough that they perform pretty well.

Like any other manufactured chip, the wiring that connects the quantum dots is locked into place during the chip’s manufacture. Since different error correction schemes require different connections among the qubits, this forces us to commit to specific error-correction schemes during manufacturing. If a better scheme is developed after a chip is made, it’s probably not possible to switch to it. Less complex algorithms may benefit from simpler error-correction schemes that require less overhead, but we wouldn’t be able to switch schemes with these chips.

So, quantum dots appear to typify the trade-offs that we’re facing with quantum computing: it’s easier for us to make lots of quantum dots and all the hardware needed to manipulate them, but it’s seemingly not possible for them to benefit from the flexibility that other types of qubits have.

The whole point of this new paper is to show that this isn’t necessarily true.

Moveable dots

The new work was done in collaboration between researchers at Delft University of Technology and the startup QuTech. The team built a chip that had a linear array of quantum dots, and they started out with single electron spins at each end. Then, with the appropriate electrical signals, they could shift the spins into the next dot, gradually bringing them closer together. (And, by gradually, we mean a fraction of a second here, but relatively slowly compared to basic switching in electronics.)

Once the electrons were close enough, the spin wavefunctions overlapped, allowing the researchers to perform two-qubit gates on them. These manipulations can be used to entangle the two spins and are thus needed to build error-corrected logical qubits; these gates are also needed for performing calculations.

The researchers then confirmed that they could move the electrons back to their starting positions, after which measurements confirmed that their spins were entangled. And since quantum teleportation also requires a two-qubit gate, they showed that the process could be used for teleportation. Teleportation can enhance the sort of mobility provided by moving the qubits around, since it can be used to move states around after the qubits have been widely separated.

(Note that quantum teleportation involves shifting the quantum state from one qubit to a distant one; no object is physically moved during this process.)

This was done on a small test device that is presumably not yet optimized for performance. But the operations were done with pretty reasonable fidelity. The two-qubit gates were executed successfully over 99 percent of the time, while teleportation succeeded about 87 percent of the time. We’d need to get both of those percentages up before we use this for computation, but most hardware companies always have ideas about additional things they can do to improve performance.

On the dot

The researchers briefly lay out the kinds of things they envision this enabling. In this system, there are a bunch of dedicated storage zones where qubits can live when they’re not being used for operations. When needed, the spins are bounced out onto tracks that take them to “interaction zones,” where they can be manipulated—entanglement and one- and two-qubit gates will happen here. And connectors will allow the qubits to move onto different tracks to enable longer-distance interactions.

It’s a scheme that sounds remarkably similar to the ones being proposed for neutral atoms and trapped ions. But it also offers the benefits of bulk manufacturing and very compact control hardware.

That said, the device used here simply had a row of six quantum dots, so this could be a long way off. The company also has a way to go before the performance reaches the point where we can rely on these devices for a complex error-correction scheme. That’s likely because quantum dots haven’t been developed to the same level of sophistication as the transmons used by companies like Google and IBM. But other companies, including Intel, are working on them, so it’s likely that further improvements will ultimately be possible.

Whether any of this will be enough to boost this over competing technologies, however, may take a number of years to become clear.

Nature, 2026. DOI: 10.1038/s41586-026-10423-9 (About DOIs).