Two Quantum Breakthroughs in One Week: IBM Goes Modular, Japan Goes Neutral
Quantum computing had a busy week. On August 19, IBM announced it had connected its first modular cryogenic systems, a step toward building a quantum computer with over 1,000 qubits. Five days later, Japan’s Institute for Molecular Science revealed that “Shunkai,” the country’s first operational neutral-atom quantum computer, is now running.
These aren’t incremental press releases. They represent two different approaches to the same problem: how to build a quantum computer that actually works at scale.
IBM’s modular approach
The core challenge in quantum computing isn’t just making better qubits. It’s fitting enough of them into a system that can keep them cold, stable, and connected. Current quantum systems are limited by the physical space inside their cryogenic cooling units — the ultra-cold environments where quantum processors operate. IBM’s new modular cryogenic architecture addresses this by giving engineers 12 times more wiring space than the company’s most widely used current quantum systems.
The design is straightforward. Box-shaped cryogenic units sit side by side and connect quantum processors using IBM’s “L-coupler” technology, which allows separate quantum chips to exchange information and operate as parts of a larger system. Think of it like linking multiple computers into a cluster, except each node needs to be cooled to near absolute zero and shielded from electromagnetic interference.
The 12x improvement in wiring space is significant because wiring is one of the physical bottlenecks in quantum system design. Each qubit needs control lines to send microwave pulses and readout lines to measure its state. As qubit counts increase, the number of wires grows proportionally, and the cryogenic units need to accommodate them without introducing thermal noise or signal interference. IBM’s new architecture addresses this by redesigning the internal layout of the cryogenic modules, creating more room for the dense wiring that large-scale quantum systems require.
The L-coupler technology that connects the modules is equally important. Quantum processors in separate modules need to share quantum information without losing the fragile quantum states that make computation possible. The coupler creates a controlled quantum channel between processors, allowing them to entangle and exchange information while maintaining the isolation needed for each processor to operate correctly. This is technically one of the hardest problems in modular quantum computing, and IBM’s ability to demonstrate it in a physical system is a meaningful step forward.
IBM plans to install its Quantum Nighthawk processors in the new cryogenic modules later this year as part of expanded testing. By 2027, the company intends to use this technology to link several processors into a quantum computer with at least 1,000 programmable qubits. That system, called Starling, represents IBM’s path toward fault-tolerant quantum computing — a state where quantum calculations can run reliably despite the errors that naturally occur in quantum systems.
The modular approach matters because it solves a scaling problem that monolithic designs can’t. Building a single cryogenic unit large enough to house thousands of qubits requires engineering solutions that don’t exist yet. Building smaller units and connecting them is harder than it sounds — quantum entanglement between processors in separate modules is extremely delicate — but it’s a more realistic path to scale.
Japan’s neutral-atom milestone
While IBM works on connecting processors, Japan’s Shunkai quantum computer takes a different approach entirely. Instead of superconducting qubits (which IBM uses), Shunkai uses neutral atoms as qubits, captured in an array using optical tweezers — tightly focused laser beams that hold individual atoms in place.
The system is “full-stack,” meaning it integrates all the layers needed to convert user inputs into drive signals for the computing device and produce computational results. Quantum calculations are performed by irradiating the atoms with microwaves or laser light. The computational results are interpreted by observing the fluorescence from each individual atom with a camera.
The optical tweezer technology that Shunkai uses is particularly elegant. A tightly focused laser beam creates a small potential well that can trap a single atom. By arranging many of these traps in a grid pattern, researchers can create a precisely controlled array of qubits. The atoms hover in place, held by light, and quantum operations are performed by hitting them with carefully tuned laser pulses or microwave signals.
What makes the full-stack integration notable is that it means Shunkai isn’t just a laboratory demonstration of a few qubits. It’s a complete system that can accept computational problems, translate them into quantum operations, execute them on the atom array, and produce results. That transition from “we can trap atoms” to “we can compute with trapped atoms” is where many quantum computing projects stall, and Japan has now crossed that threshold.
Shunkai was developed by a research team led by Professor Kenji Ohmori at the Institute for Molecular Science, part of Japan’s National Institutes of Natural Sciences. Infleqtion, a quantum computing company based in Colorado, supplied the quantum processing unit. The project is part of Japan’s Quantum Moonshot program, a government-backed initiative to develop fault-tolerant quantum computers.
Neutral-atom quantum computing has been gaining momentum because it offers several advantages over superconducting approaches. Atoms are identical by nature — every qubit of the same isotope is exactly the same, which eliminates a manufacturing variability problem that superconducting qubits face. Neutral-atom systems can also be reconfigured dynamically, with atoms moved and rearranged during computation using the optical tweezers.
Another advantage is coherence time. Neutral atoms can maintain their quantum states for relatively long periods compared to superconducting qubits, which need to be operated extremely quickly before decoherence degrades the computation. Longer coherence times mean more operations can be performed before errors accumulate, which reduces the burden on error correction systems.
The tradeoff is speed. Superconducting qubits can be operated at microwave frequencies, allowing very fast quantum gate operations. Neutral-atom systems typically use laser pulses, which are slower but offer more flexibility in terms of qubit connectivity and reconfigurability. Different applications may favor different approaches, which is why the coexistence of multiple qubit technologies is likely to continue rather than converge on a single winner.
Why both approaches matter
The quantum computing industry is often framed as a race between companies — IBM versus Google versus IonQ versus Quantinuum. But the reality is more nuanced. Different qubit technologies have different strengths, and the field is large enough that multiple approaches can succeed simultaneously.
IBM’s superconducting approach benefits from decades of semiconductor manufacturing expertise. The company has a clear roadmap (Starling, then beyond) and the industrial infrastructure to build at scale. The modular cryogenic architecture is a practical solution to a physical constraint, and it builds on IBM’s existing hardware ecosystem.
Japan’s neutral-atom approach benefits from the inherent properties of atoms themselves. The ability to reconfigure qubit arrangements on the fly gives researchers flexibility that fixed-architecture systems don’t offer. And the full-stack integration that Shunkai demonstrates — from optical trapping to fluorescence detection — shows that neutral-atom systems can be built as complete, operational computers rather than laboratory experiments.
The fact that both milestones happened in the same week is coincidental, but it illustrates a broader truth about quantum computing in 2026: the field is no longer about proving that quantum computers can work. It’s about engineering them to work reliably, at scale, for problems that classical computers can’t solve.
The neutral-atom approach also has a practical advantage in qubit connectivity. In superconducting systems, qubits are fixed in place on a chip, and interactions are limited to nearest neighbors. In neutral-atom systems, atoms can be physically moved during computation, allowing any qubit to interact with any other. This flexibility makes certain types of quantum algorithms more efficient to implement and reduces the overhead needed for error correction.
For IBM, the modular approach addresses a different constraint. Superconducting qubits are manufactured on chips using processes similar to semiconductor fabrication. The challenge isn’t making individual chips — it’s making them large enough and connecting them reliably. The modular cryogenic system is IBM’s answer to that constraint, and it’s a bet that the path to thousands of qubits runs through connecting many smaller processors rather than building one massive one.
What comes next
IBM’s Nighthawk processor installations later this year will be closely watched. If the modular cryogenic system can maintain quantum coherence across connected processors, it validates the architectural approach and clears a path toward Starling’s 1,000-qubit target. If it can’t, IBM will need to revisit the engineering. The company has been transparent about its roadmap, publishing detailed specifications and timelines, which makes it easier for the broader industry to track progress — and harder for IBM to quietly move goalposts.
The error-correction aspect of IBM’s work deserves particular attention. Quantum error correction is the process of protecting quantum information from the noise and decoherence that naturally occur in quantum systems. IBM has reported progress on a new error-correction code that significantly reduces the number of physical resources required for fault-tolerant quantum computing. If that code works as promised in conjunction with the modular cryogenic architecture, it could accelerate the timeline for practical quantum computing by reducing the qubit overhead needed for error correction.
For Japan, the immediate next step is likely scaling Shunkai to more qubits and running increasingly complex algorithms. The Quantum Moonshot program has a clear mandate: build fault-tolerant quantum computers by the 2030s. Shunkai provides the hardware foundation, and the Ohmori team will presumably push it toward larger qubit counts and longer coherence times. The involvement of Infleqtion, a company with experience in neutral-atom systems from the U.S. defense sector, also suggests that Shunkai could benefit from commercial-grade components and engineering practices that pure academic projects often lack.
Japan’s Shunkai system will likely be used for fundamental research before any commercial applications. But the fact that it’s operational — not just a prototype or a simulation — puts Japan in a small group of countries with functioning quantum computers. The Quantum Moonshot program’s goal of building fault-tolerant quantum computers by the 2030s now has a real hardware platform to build on.
For the rest of the industry, these milestones reinforce a pattern: quantum computing is progressing through engineering, not breakthroughs. The physics has been understood for decades. What’s changing is the ability to build systems that implement that physics reliably enough to be useful. That’s slower and less dramatic than a single “quantum supremacy” headline, but it’s how real technology gets built.
