A silicon quantum processor that runs itself

On July 29, 2026, HRL Laboratories published a quantum computing milestone in Nature that slices through one of the field’s most stubborn bottlenecks: an 18-qubit silicon quantum processor that runs itself. The processor is paired with a custom CMOS controller — a chip designed and fabricated in-house — that sits inside the cryostat at -450°F, performing error correction autonomously without any real-time connection to room-temperature electronics. It is, in a practical sense, a self-operating quantum computer, and it eliminates the sprawling racks of wiring and control hardware that have made scaling these machines an exercise in managing a cable nightmare.

That HRL paper is one of four major advances in spin qubit technology published in the same issue of Nature. The other groups — at QuTech in the Netherlands, the University of New South Wales, and a Japanese collaboration led by RIKEN — each reported their own progress, but the results cluster around a shared signal: spin qubits, long regarded as the underdog of quantum computing, are suddenly moving fast. QuTech’s team achieved a single-qubit error rate of 0.3% with a silicon-based device, while HRL’s error rate came in at 0.02% — a difference that reflects the gap between a well-engineered lab demonstration and a system architected from the start for autonomous operation. Neither number is a rounding error; both are solid enough to start building error-corrected logical qubits on top of them.

The wiring problem these papers address is real and has been a drag on the entire field. A conventional quantum processor requires a dedicated coaxial cable for every qubit, running from room temperature down to the millikelvin stage, and each cable introduces thermal noise, physical bulk, and a hard ceiling on how many qubits you can address before the cryostat simply runs out of space. HRL’s approach — integrating the controller directly into the cold environment and letting it handle gate operations, readout, and error correction on its own — is not a new idea, but it is one that has been genuinely difficult to execute, and the Nature paper is the first to show it working at a scale that matters.

What makes this more than a neat engineering trick is that it shifts the scaling conversation from “how many qubits can we fit in a fridge” to “how many qubits can we lay out on a chip.” Spin qubits in silicon are fabricated using the same processes that produce conventional semiconductor chips, which means the industry already knows how to make them at density. The controller chip HRL built sits alongside the qubit array inside the same cryostat, and the two communicate directly — a closed-loop system that doesn’t need to phone home to a rack of electronics for every operation. Autonomy here is not a buzzword; it’s the architectural difference between a quantum processor that can actually run error correction and one that drowns in its own control overhead.

Nature’s decision to publish four papers on the same day is its own kind of signal. Spin qubits have spent years in the shadow of superconducting qubits, which have been the workhorse of Google’s and IBM’s flagship machines. The knock on spin qubits has always been that they’re harder to control and slower to entangle, and the knock on silicon quantum dots specifically is that they’re sensitive to defects at the atomic scale. The work published Tuesday doesn’t erase those challenges, but it shows that the control problem is solvable with integrated CMOS and that the error rates are now low enough to make error correction a credible near-term goal rather than a theoretical aspiration.

None of this means a useful fault-tolerant quantum computer arrives next year. But the four papers together represent a genuine inflection point: the moment spin qubits stopped being a promising physics experiment and started behaving like a platform that can be engineered, integrated, and scaled. HRL’s autonomous processor is the most dramatic embodiment of that shift — a chip that runs itself, in the dark, at a temperature colder than deep space, correcting its own mistakes without asking anyone for help.

Sources

• Underdog ‘spin qubits’ leap forward in race to a useful quantum computer — Nature
• HRL demonstrates a silicon quantum processor that runs itself — HRL Laboratories