IBM just bought a quantum bet it doesn’t build. Days later, the payoff hit Nature.

Six days ago, IBM agreed to buy a quantum-computing lab that builds a kind of qubit IBM does not. This week, the lab showed the world why.

HRL Laboratories published a result in Nature on Wednesday: a silicon quantum processor that, in effect, runs itself. The Malibu research house is jointly owned by Boeing and General Motors. IBM had agreed to acquire it on 23 July, before the paper was public.

The wiring problem

Every quantum computer hits the same scaling trap. The qubits sit in a refrigerator near absolute zero. The electronics that control them sit at room temperature, in racks. Each qubit needs its own control lines running down into the cold. Scale to the millions of qubits a useful machine will need, and the wiring becomes an impossible tangle.

HRL moved the controller into the cold. It built a custom chip that runs at about 4 kelvin, inside the fridge beside the qubits. That chip generated every control signal for an 18-qubit device. Crucially, it ran error correction on its own, with no real-time help from room-temperature electronics, the company said. HRL says that is a first.

How they kept the qubits cold

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!

Putting warm electronics next to cold qubits sounds self-defeating. Even at 4 kelvin, the controller is hot compared with the qubits below it. The fix is a new superconducting ribbon cable. It carries hundreds of control signals down to the qubits, but almost none of the heat. The fragile quantum states survive.

The numbers back the claim. HRL reports control errors ten times lower than any previous demonstration with this type of qubit. Each operation takes under a microsecond. When the team added more qubits to its error-correcting code, the error rate fell about fivefold. That is the “error suppression” every quantum computer will depend on, and the results matched the team’s models.

Manufacturable, not just powerful

The point HRL keeps making is about cost, not just performance. The qubit chip, the cable and the controller all use standard semiconductor processes. That is the whole pitch.

“The technologies that enabled conventional computing weren’t just the highest-performing, they were the ones that could be manufactured cheaply and at scale. We think quantum computing will follow a similar path. Our goal is to build these powerful computers using standard microchip production lines and fit each one inside a single refrigerator. This approach will keep production costs low and make the technology affordable enough to tackle a much wider variety of business and scientific problems.”

That is Rob Vasquez, HRL’s chief executive. His argument is simple. The winning quantum machine will be the one you can actually build, the way silicon won conventional computing.

Why IBM wanted it

Here is where the acquisition fits. IBM has bet its quantum roadmap on a different qubit, the superconducting kind. Its big machine, Starling, is due in 2029. HRL’s silicon spin qubits are a rival approach. Buying HRL is a hedge, and more.

Both qubit types use silicon, so HRL’s control chips, interconnects and packaging help IBM whichever one wins. IBM’s research chief Jay Gambetta said the HRL team “brings a broad portfolio of technologies that will strengthen IBM’s long-term plans.”

It also gives IBM a second horse in a race it is used to leading, at a moment when governments treat quantum as strategic. IBM did not disclose the price, and the deal should close by the end of September. Boeing and GM keep working with IBM afterwards. The timing is the tell: IBM bought the prize on 23 July, and the rest of us saw the trophy six days later.