What a Big Week in Trapped-Ion Quantum Computing Actually Means

Within a span of 48 hours during the week of June 15, 2026, two papers landed that, taken together, say something important about where trapped-ion quantum computing is heading.

• On June 15, a team at Duke University, including researchers affiliated with IonQ, posted a preprint to arXiv demonstrating the first fully distributed three-node quantum network using individual atomic memory qubits connected by photons.¹
• On June 17, Quantinuum published a peer-reviewed article in Nature describing Helios, a 98-qubit trapped-ion quantum processor that sets new records across every major performance benchmark.²

Neither paper announces that quantum computing has arrived, but together they illuminate something more useful for investors to understand:

The two leading trapped-ion companies are pursuing fundamentally different theories about how this technology scales. Figure 1 illustrates the distinction.

• Quantinuum: Think of this as a chip, and we can move the ions around the chip to accomplish different things.
• IonQ: Think of this as a network, and the entanglement of the network as a whole is what becomes useful for computation.

Figure 1: Considering, Visually, Quantinuum & IonQ’s Approaches

Quantinuum’s Bet: Build the Best Single Machine

To understand what Quantinuum has built, it helps to think about the core engineering challenge in trapped-ion quantum computing. Individual trapped ions are among the most pristine qubits in existence, which means they hold quantum states with extraordinary fidelity, they are identical to one another by the laws of physics, and they interact through the well-understood Coulomb force.

The problem is arranging many of them into a machine that can run complex computations without the whole system becoming too noisy, too slow, or too difficult to control as the qubit count grows.

Quantinuum’s answer is the quantum charge-coupled device architecture, or QCCD. The idea, which traces back to a 2002 proposal by Kielpinski, Monroe, and Wineland, is to treat a quantum processor the way classical computer architects treat a chip:³

1. Separate the memory from the logic
2. Move data between them efficiently
3. Design specialized zones for specialized tasks

In Helios, Quantinuum’s system, ions not being operated on rest in a storage ring. When they’re needed, they move through an X-shaped junction into a cache region, then into one of eight quantum logic zones where gates are applied. While gates are running in the logic zones, the next batch of ions is already cooling and sorting in parallel. The architecture is explicitly designed so that the system does not have to stop and wait.

The performance numbers in the Nature paper are, by any honest reading, impressive. Averaged across all operational zones in the system, Helios achieves average single-qubit gate infidelities of 2.5 × 10⁻⁵, two-qubit gate infidelities of 7.9 × 10⁻⁴, and state preparation and measurement errors of 3.3 × 10⁻⁴. In plain English: the elementary operations are highly accurate by today’s quantum computing standards, and they remain accurate at 98 qubits. The paper also reports system-level random circuit sampling benchmarks demonstrating that classical simulation of Helios circuits would require compute resources well beyond existing supercomputers. Quantinuum describes Helios as commercially launched and cloud-accessible. This is a peer-reviewed Nature article describing a real integrated system, not a laboratory physics demonstration with a press release attached.

Perhaps the most underappreciated element of Helios is the software. Quantinuum built a real-time classical control stack called the Helios runtime that maps user-defined “virtual qubits” to physical ions while a quantum program is actively executing, dynamically routing ions, sorting batches, and supporting complex control flow including mid-circuit measurements and conditional logic. That is a meaningfully different kind of achievement than demonstrating a beautiful two-qubit gate in isolation. It is engineering infrastructure for a programmable machine.

IonQ’s Bet: Connect Many Smaller Modules

IonQ, through the Monroe research group at Duke, is pursuing a different theory. The worry with the single-processor approach is that even the most elegant chip architecture eventually hits physical limits, for example control complexity, crosstalk, heating, and fabrication tolerances. If that’s true, the path to very large-scale quantum computing may require connecting many smaller, high-fidelity modules rather than building one enormous monolithic machine. Ions would do the computing; photons would do the connecting.

The recent preprint from the Duke group reports the first fully distributed GHZ state across a three-node quantum network of individual atomic memory qubits. Each of the three nodes holds a single barium-138 ion in its own trap, separated by about two meters. Each ion emits a photon entangled with its internal quantum state. Those photons travel through three-meter optical fibers to a central GHZ-state generator, which is a carefully arranged set of beamsplitters and photodetectors. When a successful three-photon coincidence event is detected, the physics of the interferometer heralds that the three remote ions have been projected into a maximally entangled GHZ state without any of them ever directly interacting.

Why does the specific type of entanglement matter? A GHZ state is genuine three-party quantum entanglement, not just pairwise correlations between nodes A–B and B–C, but a single shared quantum object spanning all three. The team demonstrates this by measuring a Mermin parameter of 3.203, violating the classical bound by 27 standard deviations and closing the so-called detection loophole for the first time in a fully distributed system. Trapped ions can close that loophole, where pure photonic approaches historically could not, because their state detection efficiency is near-perfect.

The result is technically elegant, but it is also important to be honest about where it sits on the road to practical quantum networking. The experiment achieved a bounded GHZ state fidelity of 0.841 to 0.881, which is genuinely impressive for a three-node distributed system, better than prior demonstrations in diamond color centers or atomic ensembles. But the entanglement generation rate is approximately 0.095 successful events per second. The population data in the paper came from 687 successes out of more than 4.1 billion attempts.

The photon collection efficiency per node sits around 1%, and because the three-node rate scales roughly as the cube of that efficiency, the raw physics of current free-space collection creates a substantial engineering gap between this demonstration and the rates that would be needed for large-scale modular quantum computing.

The paper is clear-eyed about this, and outlines paths forward, which could include sympathetic cooling to increase duty cycle, integrated optical elements, and two-photon protocols that could scale more favorably.

A Philosophical Fork, Not a Horse Race

It would be easy to read these two papers as evidence that Quantinuum is “winning” and IonQ is catching up, or vice versa. In my opinion, that framing could risk missing what is actually interesting. The two companies are not competing to solve the same problem in the same way. They are, instead, making different architectural bets about where the hardest bottlenecks will eventually appear.

• Quantinuum is betting that a sufficiently clever single-processor architecture, with good ion transport, parallelized operations, and real-time software, can scale far enough to reach fault tolerance without requiring distributed networking.
• IonQ is betting that physical limits on single-processor scaling will ultimately drive the industry toward modular architectures, and that photonic interconnects are therefore a foundational investment worth making now, even if the near-term rates are modest.

These bets are not mutually exclusive in the long run. It is entirely plausible that near-term fault-tolerant computation arrives on a machine that looks more like Helios, while large-scale distributed quantum computing eventually requires the photonic networking infrastructure IonQ is building. The field is young enough that maintaining architectural diversity is almost certainly healthy.

What any investor, in my opinion, should resist is collapsing the distinction. A 98-qubit integrated processor with peer-reviewed system benchmarks and a commercial cloud offering is a different kind of milestone than a three-node lab demonstration of distributed entanglement at sub-Hz rates.

Both are representative of genuine scientific progress; they just sit at different points on the evidence ladder from physics to product.

What This Means for the Timeline Question

The question investors most often want answered about quantum computing is a simple one, and it could be distilled down to something that feels like:

Will this work, and when?

Neither paper this week answers that question definitively. What they do, collectively, is provide evidence that the trajectory is real.

Five years ago, demonstrating a high-fidelity two-qubit gate in a single trapped-ion trap was a significant experimental achievement. Today, a trapped-ion processor runs 98 qubits with all-to-all connectivity, executes circuits beyond classical simulation, and is available to customers on the cloud.

Meanwhile, the modular networking primitives required for a distributed quantum internet have moved from theoretical proposals to laboratory demonstrations with real error budgets and real rate calculations. The bottlenecks are becoming more visible, which is precisely what has to happen before they can be engineered away.

For the skeptic asking whether quantum computing will ever produce commercial value, the honest answer remains that it is not at scale yet.

Conclusion: Reflecting this in the WisdomTree Strategy

The WisdomTree Quantum Computing Fund (WQTM) is designed to track the total return performance of, before fees, of the WisdomTree Classiq Quantum Computing Index. If we boil it down, the idea is to recognize that there are many different things going on in quantum computing, with many different types of companies making their own contributions, and we want to generate strong exposure to those firms really pushing the field forward.

¹ Goetting, I., Kalakuntla, A., Shalaev, M., Shi, H. B., Ferrari, A., Saha, S., Toh, G., Male, S. & Monroe, C. (2026). Tripartite entanglement of remote atomic qubits (arXiv:2606.17173v1). arXiv.

² Ransford, A., Allman, M. S., Arkinstall, J., Campora, J. P., III, Cooper, S. F., Delaney, R. D., Dreiling, J. M., Estey, B., Figgatt, C., Hall, A., Husain, A. A., Isanaka, A., Kennedy, C. J., Kotibhaskar, N., Madjarov, I. S., Mayer, K., Milne, A. R., Park, A. J., Reed, A. P., … Bohnet, J. G. (2026). A 98-qubit trapped-ion quantum computer with all-to-all connectivity. Nature.

³ Kielpinski, D., Monroe, C. & Wineland, D. J. (2002). Architecture for a large-scale ion-trap quantum computer. Nature, 417, 709–711.