atom computing
Today, a startup called Atom Computing announced that it is conducting internal testing of a 1,180-qubit quantum computer and plans to make it available to customers next year. The system represents a major breakthrough for the company, which had previously built only one system based on neutral atomic qubits. The system operated using just 100 qubits.
The error rate for individual qubit operations is so high that it is impossible to run algorithms that rely on the full qubit count without failing due to errors. But it supports the company’s claims that its technology is rapidly scalable and provides a testbed for quantum error correction research. The company also says that for small-scale algorithms, simply running multiple instances in parallel increases the likelihood of returning the correct answer.
Computing with atoms
Atom Computing, as its name suggests, chose neutral atoms as qubits (there are other companies working with ions). These systems rely on a series of lasers that create a series of energetically favorable positions for the atoms. If left alone, atoms tend to fall into these locations and remain there until stray gas atoms collide and destroy them.
The positions of the atoms are set by the laser configuration, so each can be specified individually. Quantum information is stored in nuclear spins, which are relatively unaffected by the environment. While other types of qubits have coherence lifetimes of just a few seconds, neutral atoms often remain in their state for tens of seconds. Because nuclear spins do not easily interact with the environment, atoms can be packed tightly together, allowing relatively dense systems.
However, it is possible to manipulate atoms so that they interact and intertwine. This works by the so-called Rydberg cutoff, which prohibits two atoms from interacting unless they are a certain distance apart and both are in the Rydberg state. In this state, the outermost electrons of the atom are loosely bound and orbit a long distance from the nucleus. By putting suitable pairs of atoms into Rydberg states (which can also be done with lasers), it is possible to entangle them. Also, since the laser allows you to control the position of individual atoms, it is possible to entangle any two atoms.
Because the system allows atoms to be packed relatively densely, Atom Computing argues that the system is well-positioned to scale quickly. Systems like Transmon ensure that all trapped atoms behave in the same way, as small differences in device fabrication lead to small variations in qubit performance. Additionally, atoms do not cause crosstalk unless manipulated, making it possible to pack a large number of atoms into a relatively small space.
These two factors mean the neutral atoms are well-positioned to scale up to large numbers of qubits, company executives argue. Its original system, which went live in 2021, was a 10×10 grid of atoms (although three-dimensional arrangements are also possible). And when I spoke to Ars a year ago, they said they wanted to scale their next-generation system by an order of magnitude, but they didn’t say when that would be ready.
