Atom-based quantum computers are catching up in the race to usefulness

A neutral-atom quantum computer built with extremely cold, electrically neutral atoms has hit a new milestone: it can run quantum error correction repeatedly, checking for mistakes up to 90 times in a row. Researchers say the result is a practical step toward

For years, the quantum computing race has been less about building machines that can run a clever demonstration—and more about whether they can keep correcting themselves when they inevitably slip up. In that fight, a neutral-atom approach has just cleared a demanding hurdle.

Atom Computing’s team. led by Ben Bloom. built what’s often called a neutral-atom quantum computer made from extremely cold atoms. The central achievement is error correction that doesn’t just work once. but can be repeated: the system kept running and checking for errors. monitoring the “alert” qubits that signal when something has gone wrong. up to 90 times in a row.

“That’s a big check mark for what you can do in a neutral-atom system,” Bloom said. “The differences between [experiments] we were doing before were big step changes, but now, it is just about building it better, faster, cheaper.”

The need is urgent because quantum computers are notoriously error-prone. Even small disturbances can derail calculations. and one of the biggest obstacles to usefulness is figuring out how to recognize a computational error. discard and restart the calculation. and do so reliably enough that the machine can actually produce useful results.

Error correction is based on spreading information across groups of qubits. In Atom Computing’s setup, some qubits act as an alert system. If those alert qubits indicate an error, the calculation can be thrown out and restarted.

The team’s tests showed they could scale up the size of those qubit groups used for error correction—from groups of 16 to groups of 32—without introducing additional errors. In fact, the error rates were lower with the larger grouping. That matters because the route to more power in a quantum computer depends on building up qubit counts and doing so without sacrificing reliability.

The wider landscape has been moving quickly. In 2023, researchers at Google simultaneously increased the qubit number and decreased the error rate in a superconducting quantum computer. A team at the University of Science and Technology of China did something similar in 2025. And in 2025, researchers at Harvard University reported the same kind of improvement for another neutral-atom quantum computer.

What Bloom’s team emphasizes as different is not only scaling the error-correction mechanism, but sustaining it. “The goal was always… to run error correction at infinitum,” he said.

Still, the bigger test is what comes after the lab wins. Solving industrially relevant problems will require both lots of qubits and computations that can reliably keep going. Atom Computing argues the latest work supports exactly that—showing the ability to combine the capabilities needed to build a real neutral-atom quantum computer in a single experiment.

Jeff Thompson at Princeton University said the new study “is the first to bring together all of the capabilities needed to build a real neutral-atom quantum computer in a single experiment.” He called the approach an experimental tour de force. while adding that there is still room to improve overall error rates and the speed of computation.

Mark Saffman at the University of Wisconsin–Madison also framed the result as part of a longer push toward a neutral-atom machine that can run continuously. more like conventional computers. But he cautioned that the promise isn’t fully intact yet: as the system kept working through those 90 rounds of checking for errors. additional errors did accumulate. That accumulation, he said, detracts from how quickly the technology can reach its “useful” threshold.

Bloom acknowledged the problem—at least in principle. He said he and his colleagues are already working on addressing some of the errors, and he expressed confidence that the system’s performance can be continuously improved.

Taken together, he argues, the momentum in neutral-atom work places the approach as a formidable competitor to other builds, including superconducting qubits.

“What this work is showcasing is that a lot of the physical mechanisms that stop neutral atoms from being as awesome as superconducting qubits are starting to disappear,” Bloom said. Thompson agreed that progress should accelerate. “I expect rapid progress to follow… across the industry,” he said.

quantum computing neutral-atom quantum computer error correction qubits Atom Computing ultracold atoms superconducting qubits quantum error correction milestones