I saw a quantum computer used by JPMorgan, with Microsoft and Nvidia as partners, and it was mind-blowing

• Quantum computers are wondrous because they operate on quantum mechanics.
• Quantinuum unveiled a new one in a Denver data center and invited a small crew of people to see it.
• The computer looked like a very sophisticated science experiment.

A drab, vinyl-floored, windowless office building just north of Denver houses one of the wonders of the computer industry. It's home to several wonders, actually: a data center where two commercial-quantum computers are operating for paying customers including JPMorgan, and a third prototype under construction.

Quantum computers are wondrous because they operate on quantum mechanics: the tiniest parts of the universe where the rules of physics as we think we know them don't always apply. Instead of transistors that calculate information by switching on and off, quantum computers use qubits, which can switch to "on," "off," or "both," meaning some state in between.

Proponents say these computers are uniquely able to solve complex problems, such as discovering new materials or determining how to improve supply chains.

Quantinuum, which spun out of the defense contractor giant Honeywell in 2021 after Honeywell acquired the UK quantum company Cambridge Quantum, built the computer I saw. I visited as part of an invitational media-and-analysts' day held a few days before Quantinuum officially launched H2, its 32-qubit, second-generation quantum computer on Tuesday.

Quantinuum claims the H2 is the most precise quantum computer ever built — and it's published the results of numerous benchmarks tests it conducted to back its bravado. 

That's a significant claim because for these computers to reach their potential and one day be far more powerful than today's supercomputers, they'll need to use hundreds, a thousand, or even more qubits. And to do that, the qubits must stay on task, calculating what engineers tell them to compute.

Infinitesimal environmental changes, such as temperature or light, can interfere with their work. The more qubits crammed together, the higher the likelihood that some will get distracted by that so-called noise, and the higher the likelihood that the computer's overall results will be less accurate.

Because of that, some skeptics have said that quantum computing at that scale may not be possible. Still, Russell Stutz, the executive responsible for the design and build of these computers at Quantinuum, told me that his work has made him a believer. "It's not if, but when," he said.

Other competitors focus on size records. In November, IBM unveiled a 400-qubit machine, and Google has used a 54-qubit processor called Sycamore since 2019. A handful of startups are working on quantum as well.

No one can say how long it will be before quantum computers reach the scale and reliability to meet their promise, but they are already in use for early adopters experimenting with the tech. Companies like Nvidia and Microsoft, which is building its own quantum computer, are partners to Quantinuum.

There's still work to be done because even if a quantum computer can claim accuracy as high as 99%, that only "ensures that you will get the wrong results" when making millions of calculations, Fabrice Frachon, the principal PM lead at Microsoft's Azure Quantum unit, told me.

All that explains why Quantinuum's computer was amazing — it looked more like a science experiment than a piece of next-generation computing technology.

The H2 is the size of a small room

We were forbidden from taking pictures of the actual computer or data center, but the company shared this basic, schematic drawing of what the H2 computer looks like. It doesn't do it justice. 

As the graphic above shows, the computer is the size of a small room, where its components are splayed across several tables instead of contained in a boxed casing, similar to a typical supercomputer.

Using the best artistic skills I have — yes, really — I've annotated the above illustration to show a bit better what the room really looked like. 

The computer was actually spread across two 100-square-foot platforms, and my stick-figure self stood in an aisle between them. Blue cables sprouted out all over the place, neatly organized, piping into conventional, but customized, computing devices that control the computer and the environment. Quantinuum uses lasers to cool and control the chamber the qubits are in.

Knobs and custom controls lay all over the lower tables. The computers are usually hidden behind black curtains that shut out the environment, and a group of Ph.D. scientists work in the room with monitors at the back against the wall. 

The room controls a thumbnail-sized tray with 32 qubits on a racetrack

Quantinuum's computer uses "trapped ions" as the foundation of its qubits, meaning that atomic scientists take atoms — from the mineral ytterbium — strip off electrons, charge them, and place them in a tiny thumbnail-sized tray that restricts their movements. 

This H2 computer is currently 32 qubits, but the company says it will grow to 50. It's building a prototype that could handle more. That version will use a grid-patterned tray, and the about-100-square-foot area for the computer won't need to be bigger, Stutz said. 

The tiny qubits are inside a metal chamber about the size of a cantaloupe that needs to be kept so cold that it's almost "absolute zero," the lowest temperature possible. The qubits emit light, which indicate if they are on, off, or in an in-between state.

As Quantinuum makes its computers more powerful, it also hopes to shrink them over time. As for when these room-sized computers could shrink in size and cost and become another PC on your desk, that won't happen for the foreseeable future, Stutz said. For now, this remains a cloud service, where customers send in jobs as easily as uploading a document.

Looking at the Quantinuum computer and hearing about all the science used to invent it made me feel like we are standing on the cusp of a new era, similar to when the computing pioneer Ada Lovelace explained the potential of an "analytical engine."