In a small basement laboratory, Harry Levine, a Harvard University graduate student in physics, can assemble a rudimentary computer in a fraction of a second. There isn't a processor chip in sight; his computer is powered by 51 rubidium atoms that reside in a glass cell the size of a matchbox. To create his computer, he lines up the atoms in single file, using a laser split into 51 beams. More lasers—six beams per atom—slow the atoms until they are nearly motionless. Then, with yet another set of lasers, he coaxes the atoms to interact with each other, and, in principle, perform calculations.

It's a quantum computer, which manipulates "qubits" that can encode zeroes and ones simultaneously in what's called a superposition state. If scaled up, it might vastly outperform conventional computers at certain tasks. But in the world of quantum computing, Levine's device is somewhat unusual. In the race to build a practical quantum device, investment has largely gone to qubits that can be built on silicon, such as tiny circuits of superconducting wire and small semiconductors structures known as quantum dots. Now, two recent studies have demonstrated the promise of the qubits Levine works with: neutral atoms. In one study, a group including Levine showed a quantum logic gate made of two neutral atoms could work with far fewer errors than ever before. And in another, researchers built 3D structures of carefully arranged atoms, showing that more qubits can be packed into a small space by taking advantage of the third dimension.

The advances, along with the arrival of venture capital funding, suggest neutral atoms could be on the upswing, says Dana Anderson, CEO of ColdQuanta, a Boulder, Colorado–based company that is developing an atom-based quantum computer. "We've done our homework," Anderson says. "This is really in the engineering arena now."

Because neutral atoms lack electric charge and interact reluctantly with other atoms, they would seem to make poor qubits. But by using specifically timed laser pulses, physicists can excite an atom's outermost electron and move it away from the nucleus, inflating the atom to billions of times its usual size. Once in this so-called Rydberg state, the atom behaves more like an ion, interacting electromagnetically with neighboring atoms and preventing them from becoming Rydberg atoms themselves.

Physicists can exploit that behavior to create entanglement—the quantum state of interdependence needed to perform a computation. If two adjacent atoms are excited into superposition, where both are partially in a Rydberg state and partially in their ground state, a measurement will collapse the atoms to one or the other state. But because only one of the atoms can be in its Rydberg state, the atoms are entangled, with the state of one depending on the state of the other.

Once entangled, neutral atoms offer some inherent advantages. Atoms need no quality control: They are by definition identical. They're much smaller than silicon-based qubits, which means, in theory, more qubits can be packed into a small space. The systems operate at room temperature, whereas superconducting qubits need to be placed inside a bulky freezer. And because neutral atoms don't interact easily, they are more immune to outside noise and can hold onto quantum information for a relatively long time. "Neutral atoms have great potential," says Mark Saffman, a physicist at the University of Wisconsin in Madison. "From a physics perspective, [they could offer] easier scalability and ultimately better performance."

The two new studies bolster these claims. By engineering better quality lasers, Levine and his colleagues, led by physicist Mikhail Lukin at Harvard, were able to accurately program a two-rubidium atom logic gate 97% of the time, they report in a paper published on 20 September in Physical Review Letters. That puts the method closer to the performance of superconducting qubits, which already achieve fidelity rates above 99%. In a second study, published in Nature on 5 September, Antoine Browaeys of the Charles Fabry Laboratory near Paris and his colleagues demonstrated an unprecedented level of control over a 3D array of 72 atoms. To show off their control, they even arranged the atoms into the shape of the Eiffel Tower. Another popular qubit type, ions, are comparably small. But they can't be stacked this densely because they repel each other, acknowledges Crystal Senko, a physicist at the University of Waterloo in Canada who works on ion quantum computers.

Not everyone is convinced. Compared with other qubits, neutral atoms tend not to stay put, says Varun Vaidya, a physicist at Xanadu, a quantum computing company in Toronto, Canada, that builds quantum devices with photon qubits. "The biggest issue is just holding onto the atoms," he says. If an atom falls out of place, Lukin's automated laser system can reassemble the atoms in less than a second, but Vaidya says this may still prohibit the devices from performing longer tasks. "Right now, nobody knows what's going to be the best qubit," Senko says. "The bottom line is, they all have their problems."

Still, ColdQuanta has recently received .75 million in venture funding. Another startup, Atom Computing, based in Berkeley, California, has raised million. CEO Ben Bloom says the company will pursue qubits made of atoms with two valence electrons instead of rubidium's one, such as calcium and strontium. Bloom believes these atoms will allow for longer-lived qubits. Lukin says he's also interested in commercializing his group's technology.

The startups, as well as Saffman's group, are aiming to build fully programmable quantum computers. For now, Lukin wants his group to focus on building quantum simulators, a more limited kind of computer that specializes in solving specific optimization problems by preparing the qubits a certain way and letting them evolve naturally. Levine says his group's device could, for example, help telecommunications engineers figure out where to put radio towers to minimize cost and maximize coverage. "We're going to try to do something useful with these devices," Levine says. "People still don't know yet what quantum systems can do."

In the next year or two, he and his colleagues think neutral atom devices could deliver an answer.