QuEra Computing just solved one of neutral-atom quantum computing’s most tedious problems. Its new QuEra laser automation system uses Anthropic’s Claude to catch and fix laser drift on its own, in seconds. A human specialist used to need five to ten minutes for the same fix, sometimes at 2 a.m.
Why Laser Drift Slows Down Quantum Computers
Neutral-atom quantum computers trap and steer atoms with tightly tuned lasers, the same approach behind the machine Japan switched on earlier this year, its first neutral-atom quantum computer. Those lasers act as the control knobs for every qubit. But heat, vibration, and ordinary lab noise constantly push them out of tune.
When a laser drifts too far, the whole machine halts. Someone has to walk over, diagnose the fault, and manually restore the lock before the computer can run again. Newer QuEra machines pack in more lasers per generation, so this problem multiplies as the hardware scales. For a quantum computer at a customer’s site, that means flying in a specialist every time a laser slips.
Picture David, a 52-year-old facilities manager at a regional hospital piloting quantum-assisted drug interaction research. Under the old setup, a laser fault at 2 a.m. meant paging a specialist and waiting up to ten minutes with the machine sitting idle. Under the new one, the system catches the fault, fixes it, and gets back to work in under six seconds, and nobody’s pager ever has to go off.
Inside the QuEra Laser Automation Fix
QuEra engineers didn’t write a rulebook for every possible fault. Instead, they handed Claude the problem, the safety limits, and the target outcome. They used Anthropic’s Model Hardware Standard, a framework built with HHMI Janelia Research Campus that lets AI agents run experiments on real lab equipment safely.
Claude then ran its own experiments on a dedicated laser testbed. It proposed a fix, tested it, checked the result, and tried again, including overnight and unsupervised. It worked through hundreds of failure scenarios this way. The output wasn’t a black box, either. Claude produced ordinary, readable software, and QuEra’s engineers reviewed and approved it before it touched production hardware, as The Quantum Insider reported.
The Numbers Behind the Fix
The results held up under real lab conditions, not just clean simulations. Across seven different fault types, the system restored 695 out of 700 locks, a 99.3% success rate, without a single false alarm. Most fixes landed in under six seconds. The hardest cases took 10 to 14 seconds, still far faster than the 5 to 10 minutes a specialist needed before.
The automated tuning also left the lasers steadier than manual adjustment did: residual noise dropped fivefold. Claude even commissioned a second laser wavelength overnight, a task that used to take QuEra’s team weeks of hands-on setup. That kind of speedup echoes what’s happening elsewhere in the stack, where AI decoders are already outpacing standard methods at catching qubit errors in real time.
Why This Changes the Math on Scaling Quantum Computers
QuEra’s VP of quantum systems, Sergio Cantu, put it plainly: the company is building quantum computers that fix themselves. It no longer needs an engineer driving in at 2 a.m. That shift matters more than it sounds. A quantum computer needing a resident laser expert isn’t a product a hospital, bank, or research lab can host on-site. One that heals its own faults in seconds is.
QuEra says the same approach could extend to other subsystems beyond lasers. The company is also targeting cloud delivery of its Libra quantum system through Amazon Braket by 2028. Removing specialist dependency from even one subsystem cuts operating costs and shortens the path from lab prototype to a machine that runs unattended in a customer’s data center.