Neutral-Atom Quantum Computer Achieves Repeatable Error Correction

A quantum computer built from extremely cold, neutral atoms has achieved a crucial milestone in the race toward practical application: repeatable error correction. Ben Bloom at Atom Computing and his colleagues have constructed a system capable of repeatedly identifying and correcting computational errors, distinguishing it from previous experiments that demonstrated larger, but less consistent, improvements. This development positions neutral-atom technology as a growing competitor to the superconducting circuits favored by industry leaders Google and IBM, both of which have dominated quantum computing for a decade. “This is a big check mark for what you can do in a neutral-atom system,” says Bloom, adding that the team is now focused on “building it better, faster, cheaper.” The ability to sustain reliable computation is essential, as sufficiently powerful quantum computers promise breakthroughs in materials science and drug discovery, while simultaneously posing a threat to current internet encryption.

Neutral-Atom Systems Advance in Quantum Error Correction

Neutral-atom quantum computing is rapidly establishing itself as a viable competitor to superconducting systems in the pursuit of a fault-tolerant quantum computer. This achievement signifies substantial progress toward building a machine that can sustain complex calculations without succumbing to the inherent instability of quantum states. The team focused on improving error correction, specifically the ability of a quantum computer to recognize computational errors and restart calculations; quantum computers are notoriously prone to errors, making this a major hurdle. They successfully increased the size of qubit groupings used for error correction from 16 to 32, without increasing error rates. In fact, larger groupings exhibited lower error rates, a crucial step as increasing qubit numbers directly correlates with computational power. “The goal was always to run error correction indefinitely,” Bloom stated.

Researchers at Google, IBM, and the University of Science and Technology of China have also demonstrated simultaneous increases in qubit numbers and decreases in error rates, but Atom Computing’s sustained error checking represents a significant leap forward. Jeff Thompson at Princeton University acknowledges the experimental complexity, stating, “This study is the first to bring together all of the capabilities needed to build a real neutral-atom quantum computer in a single experiment.”

Atom Computing’s Qubit Scaling to Groups of 32

Unlike previous experiments that showed improvements in either qubit count or error reduction, the Atom Computing team, led by Ben Bloom, focused on sustained error correction, repeatedly identifying and resolving computational errors during long calculations. This is crucial because quantum computers are inherently prone to errors, and reliable correction is a primary obstacle to practical application. Despite some error accumulation over the 90 correction cycles, Bloom remains optimistic, stating his team is actively addressing these issues and confident in their ability to enhance performance. He believes this progress demonstrates that many of the limitations previously hindering neutral-atom systems are beginning to diminish.

Competitive Progress with Superconducting and Neutral-Atom Approaches

Atom Computing’s recent demonstration of sustained error correction places neutral-atom quantum computing in direct competition with the decade-long lead established by Google and IBM’s superconducting qubit systems. While superconducting approaches currently dominate the field, Ben Bloom and his colleagues have achieved a critical milestone: their neutral-atom computer can repeatedly identify and correct errors during computations, a feat previously demonstrated with less consistency. Similarly, a Harvard University team recently demonstrated improvements in another neutral-atom quantum computer. However, the Atom Computing experiment distinguishes itself through its endurance, maintaining error checking across 90 consecutive computational cycles.