Physicists built an 18-qubit silicon quantum chip that runs on its own inside a deep freeze

A team of physicists has demonstrated an 18-qubit quantum processor built on silicon that manages its own operation from inside a cryogenic chamber, without the tangle of room-temperature wiring that normally connects a quantum chip to its control electronics. The result, described in a research paper published in mid-2026, tackles one of the least glamorous but most stubborn obstacles standing between today’s experimental machines and the large, practical quantum computers researchers envision.

The advance is not about setting a record for raw qubit count, which remains modest, but about architecture. By placing the control hardware next to the qubits in the extreme cold and letting the system run error correction on its own, the design points toward a way to scale quantum processors without drowning them in cables.

The wiring problem this chip solves

In a typical quantum computer, delicate qubits sit at the bottom of a refrigerator chilled to near absolute zero, while the electronics that control and read them sit at room temperature outside. Each qubit needs its own control lines threading through the cold apparatus, and as the number of qubits grows, so does the thicket of wires, until the sheer volume of cabling becomes a physical and thermal bottleneck. That mismatch between a growing quantum chip and its ballooning wiring harness is one reason scaling has proven so hard.

The team’s answer was to move the control electronics into the cold with the qubits. As detailed by the laboratory that built the system, a custom control chip fabricated in conventional semiconductor technology was placed inside the cryostat, close to the qubits and operating at cryogenic temperatures. That co-location replaced racks of external instruments and the many wires they required, letting the processor manage itself rather than depending on a room-temperature command center.

Why silicon spin qubits matter

The processor uses spin qubits made in silicon, encoding quantum information in the spin state of individual electrons trapped in a silicon-based structure. This approach appeals to engineers because it borrows from the same materials and fabrication methods that already produce the world’s computer chips, raising the prospect that quantum processors could eventually be manufactured with mature semiconductor techniques rather than exotic one-off methods.

Silicon spin qubits are also extremely small compared with some competing designs, which is part of what makes packing control circuitry alongside them feasible. Building both the qubits and their controller on compatible silicon platforms is the kind of integration that industrial-scale manufacturing would demand, and demonstrating it in a working device rather than a simulation is a meaningful step.

Running error correction without help

Beyond the wiring feat, the system carried out error correction autonomously, executing a routine that will be essential for any full-scale quantum computer. Qubits are notoriously fragile, easily disturbed by heat, stray fields, and noise, so future machines will need to constantly detect and fix errors faster than they accumulate. The demonstration ran a repetition code, a foundational error-correction scheme, entirely under the on-chip controller’s direction.

The performance numbers reported alongside the demonstration underscored the point. The integrated design achieved control errors described as roughly ten times lower than previous efforts with this type of qubit, and the researchers observed error suppression improving by about fivefold as additional qubits were folded into the error-correcting code. That trend, in which adding qubits makes the encoded information more robust rather than less, is exactly the behavior a scalable, fault-tolerant machine must exhibit.

What it means for the road to useful quantum computers

Independent trackers of the field framed the work as a manufacturing blueprint rather than a finished product. Commentary from quantum industry analysts noted that co-integrating qubits with cryogenic control electronics addresses a scaling barrier that has loomed over every hardware platform, and that a demonstration on a silicon foundation carries added weight given the industry’s existing ability to mass-produce silicon chips.

None of this means a practical quantum computer is imminent. Eighteen qubits remain far short of the thousands or millions that useful, error-corrected machines are expected to require, and moving from a laboratory demonstration to reliable, reproducible manufacturing is a long road. But by showing that a silicon quantum chip can run itself inside the deep freeze and correct its own errors as it grows, the work sketches a plausible path for how such machines might eventually be scaled up rather than merely imagined.

Why the deep freeze is unavoidable

The extreme cold at the heart of the demonstration is not a design flourish but a physical necessity. Quantum information is stored in fragile states that are easily scrambled by heat, since thermal energy jostles particles and destroys the delicate coherence a qubit depends on. Chilling the hardware to a whisker above absolute zero quiets that thermal noise enough for the quantum states to survive long enough to be useful, which is why quantum processors of many kinds live inside elaborate refrigerators.

Placing conventional control electronics into that same frigid environment is harder than it sounds, because ordinary chips generate heat and can behave differently at cryogenic temperatures. Engineering a controller that works reliably in the cold, without dumping so much heat that it warms the qubits it is meant to serve, is a significant part of what made the demonstration notable. Solving that balance is a prerequisite for any future machine that hopes to keep its control hardware close to its qubits.

How this fits the wider race in quantum hardware

The silicon spin approach is one of several competing bets in a field that also includes superconducting circuits, trapped ions, and other qubit designs, each with distinct strengths and weaknesses. What unites the contenders is the shared challenge of scaling: a laboratory device with a handful of qubits is a very different thing from a fault-tolerant machine with the vast numbers needed to run useful algorithms. Demonstrations that tackle the plumbing of control and error correction, rather than merely adding qubits, speak directly to that scaling problem.

By showing that a silicon processor can integrate its own cryogenic control and suppress errors as it grows, the work offers a template that could carry over as qubit counts rise. Whether this particular design ultimately wins out is far from settled, and rival platforms are advancing on their own timelines. But the result narrows the gap between today’s experimental chips and the manufacturable, self-correcting processors that a practical quantum computer will require.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview