A trapped ion quantum computer called Helios just became the biggest of its kind, and the architecture is the real story

[Depot]

A guest piece published in The Conversation this week by physics professor Animesh Datta lays out why Quantinuum's new Helios quantum computer represents a genuinely significant engineering advance, and it's worth understanding exactly what makes this machine different rather than just registering another qubit count headline. Helios is a trapped ion quantum computer, meaning it works by knocking an electron off individual atoms to create charged ions that get suspended in place using electromagnetic fields, and it operates using 98 qubits, making it the largest trapped ion quantum computer built so far, up from Quantinuum's own earlier machines that ran 32 qubits in 2023 and 56 qubits in 2025.

What actually sets Helios apart isn't just the raw qubit count though, it's the underlying architecture called QCCD, short for quantum charge coupled device, which was first invented back in 2002 but is only now being pushed to this kind of scale. The design uses separate physical regions for storing qubits and for actually processing them, similar in spirit to how a classical computer separates its hard drive storage from its processor, and Helios specifically arranges this in a rosette shape with a ring for storage and two streamers feeding into processing regions, all meeting at a crucial four way X junction where the ions get physically transported back and forth using precise electric voltages.

The article explains three real advantages this approach brings to quantum computation. Gates operating on one qubit don't spill over and affect neighboring qubits the way they can in some other architectures, which reduces crosstalk and improves overall quality. Qubits can be measured and reset mid computation, letting researchers catch and correct errors at the earliest possible point rather than discovering them only after a full computation finishes. And because ions can be shuttled to very different parts of the storage memory and back, even physically distant qubits can end up connected to each other in ways that make computations more efficient than architectures where qubit position is fixed.

The genuinely new piece of engineering here is that four way X junction, which lets the system handle multiple transport and processing tasks simultaneously instead of being limited to moving data back and forth along a single line or loop the way earlier QCCD machines were restricted to. That hardware advance gets paired with new software called Helios runtime, which plans the fastest and smartest route for moving ions and processing data through that two dimensional junction space. Together those two pieces let Helios run computations that would be genuinely impractical on even the largest classical supercomputers using known methods, at least for the specific kinds of random benchmark tests it's been run on so far.

Datta is careful to keep the achievement in proper perspective though, and this is honestly the most useful part of the whole piece. Helios' computations to date have just been random benchmarks rather than anything of direct scientific or commercial value, and the longest computation it's actually run involved roughly 4,000 operations across those 98 qubits. For comparison, the UK's own National Quantum Strategy envisions a fault tolerant quantum computer eventually running something on the order of a trillion operations, which gives you a real sense of just how enormous the remaining gap is between an impressive engineering demonstration and something that can solve genuinely useful problems in science or industry

The global race to make a practical quantum computer just took a big leap forwardtheconversation.com

[Paul73]

The four way X junction is genuinely the clever part of this whole story and it doesn't get nearly enough attention outside of pieces written by an actual physicist like this one. Earlier trapped ion systems being stuck moving data along a single line or loop was a real structural limitation, and letting the system handle multiple transport tasks at once in two dimensions is the kind of architectural unlock that tends to matter more long term than just adding more qubits to an existing bottlenecked design.