That distinction matters to enterprise technology teams. Quantum computing is often discussed as a single market, yet it contains very different activities: building physical processors, operating remote machines, developing algorithms, providing optimization software, and integrating hybrid workflows with conventional high-performance computing. NEC’s documented commercial offering still includes quantum-inspired optimization software. Its dedicated Vector Annealing cloud service, however, ended in March 2025. For organizations assessing suppliers, the practical issue is therefore not whether NEC is “still in quantum” in the broadest sense, but what products and delivery models remain available now.
What the reporting establishes—and what it does not
Jiji Press reported on September 7, 2026 that NEC had discontinued quantum-computer hardware development at the end of March, attributing the information to people familiar with the matter. Diamond Online separately reported that NEC had withdrawn from quantum-computer machine development at the same point. The overlap between those accounts gives the underlying event meaningful weight.
Jiji further reported that NEC apparently concluded continued hardware work would not produce the expected profits. The reported reasoning reflects the hard economics of the field: practical deployment remains challenging, and no hardware approach has conclusively emerged as the dominant route to useful, large-scale quantum computation. Developing physical quantum systems requires sustained investment in specialized devices, control systems, software, facilities, and talent, without a guaranteed timetable for commercial returns.
Still, readers should be careful with stronger formulations. The available material does not include a direct NEC announcement confirming the reported exit. It likewise does not establish that NEC formally ended only a gate-based superconducting quantum-processing-unit program. One account used that more specific description, but the independent reports describe the discontinued activity more generally as hardware or machine development.
That difference is more than semantics. In 2023, NEC described its cooperative laboratory with Japan’s National Institute of Advanced Industrial Science and Technology as researching both gate-based quantum computers and quantum annealing. NEC and Tohoku University also announced joint research involving an eight-qubit quantum annealing machine developed by NEC and AIST. That machine used superconducting technology. Without a public program-closure list, it is not safe to infer exactly which lines were ended, wound down, transferred, or retained as research collaborations.
The reported decision may well represent a broad departure from in-house machine development. The available evidence simply cannot confirm the precise boundary of that departure.
A notable retreat for a superconducting research pioneer
NEC’s place in the history of superconducting quantum technology makes the reported move especially striking. A 1999 Nature paper by researchers affiliated with NEC Fundamental Research Laboratories in Tsukuba described coherent control of macroscopic quantum states in a single-Cooper-pair box, a candidate qubit. Jiji characterizes NEC’s 1999 achievement as the world’s first superconducting qubit.
Historical importance does not automatically translate into a commercially sustainable hardware business decades later. The distance between demonstrating a coherent quantum state and operating an economically viable computing platform is enormous. Hardware developers must advance error rates, scaling, calibration, fabrication consistency, system control, and the software tooling needed to turn a physical machine into a useful service. They must do so while prospective customers can frequently meet near-term optimization and simulation needs through conventional infrastructure.
This is the central counterweight to treating the reported exit as an indictment of quantum research itself. A company can possess meaningful scientific heritage and technical capability while deciding that building and commercializing full machines is not the most viable use of capital. Conversely, reducing the number of companies pursuing hardware can narrow domestic expertise and diminish competition in a strategically important technology area. Both realities can be true at once.
Diamond Online reported that many researchers involved in NEC’s development moved to Fujitsu. That is potentially important for Japan’s talent map, but it should remain an attributed report rather than a settled account of the number of people involved, the timing of their moves, or whether each move was caused by the reported hardware exit.
Fujitsu and Japan’s evolving hardware landscape
Fujitsu remains a major Japanese participant in superconducting quantum computing. In collaboration with RIKEN, it has developed a 256-qubit superconducting quantum computer and offers a hybrid quantum computing platform. This makes Fujitsu an obvious focal point for continuity of Japanese corporate superconducting-hardware expertise, particularly if the reported talent movement is borne out.
It would be an overreach, though, to call Fujitsu the only remaining domestic corporate hardware developer based on the available evidence. The quantum ecosystem includes corporate vendors, government-supported research institutions, universities, international providers, and hybrid computing initiatives. A single corporate exit does not reduce that landscape to one company.
Japan also has access to several distinct systems, which should not be conflated.
RIKEN says that the superconducting IBM Quantum System Two called ibm_kobe was introduced at its Kobe campus in June 2025. IBM, separately, says an IBM Quantum System One is operational at the Kawasaki Business Incubation Center. Describing System Two as being in both Kobe and Kawasaki combines two different IBM installations and produces an inaccurate picture of the infrastructure.
There is also Quantinuum’s Reimei system, a Japan-based regional deployment that continues to execute customer jobs. RIKEN says it was introduced at its Wako campus in February 2025. Reimei uses trapped-ion technology, not superconducting hardware. For users and policymakers, that distinction is important: “quantum access in Japan” is not a statement about one platform, one vendor, or one underlying engineering approach.
The resulting picture is not a binary story of Japan gaining or losing quantum capability. Rather, it is a redistribution among different institutions and modalities. That can preserve options for researchers and users, although it may change where intellectual property, systems expertise, and commercialization decisions reside.
NEC’s visible commercial position: Vector Annealing software
The clearest documented continuing NEC offering in this area is Vector Annealing, which NEC describes as a simulated or quantum-inspired annealing platform. It is aimed at combinatorial optimization: problems where a system must choose among an enormous number of possible combinations to find a high-quality solution under constraints.
Examples documented for the product include production planning, loading, and delivery-route optimization. These are tangible categories for enterprises, including organizations running Windows-based planning tools, line-of-business applications, data platforms, and hybrid cloud environments. A retailer planning shipments, a manufacturer allocating production capacity, or a logistics organization sequencing deliveries may care far more about solution quality, repeatability, integration, and total operating cost than about whether the optimization engine is a physical quantum processor.
NEC offers Vector Annealing 4.0 through software licenses for customer on-premises systems or cloud environments such as AWS. That makes the offering potentially relevant to conventional enterprise deployment patterns: an organization can evaluate it as software integrated with existing workflows rather than as an experiment requiring direct use of a remote quantum machine.
One correction is essential. NEC’s dedicated Vector Annealing cloud service built on SX-Aurora TSUBASA ended on March 31, 2025. Continuing availability of software licensing is not equivalent to continued operation of NEC’s former dedicated cloud service. Procurement teams should ask which deployment option is currently supported, where the software will run, what infrastructure and operating skills it requires, and how it connects to their existing applications and datasets.
The available evidence also does not support a claim that NEC has announced a defined post-hardware-shutdown strategy centered on logistics, finance, scheduling, or any other particular set of sectors. The documented use cases show where Vector Annealing can be applied; they do not establish a newly announced R&D-resource allocation or financial-modeling focus.
Why quantum-inspired optimization may be the pragmatic near-term bet
The appeal of quantum-inspired optimization is straightforward. Organizations often have difficult planning problems today, while fault-tolerant quantum machines capable of broad practical advantage remain an unsettled prospect. Software that runs on available classical infrastructure can be tested against incumbent solvers using real business constraints, current data, and measurable operational outcomes.
This does not mean quantum-inspired annealing is automatically better than established optimization methods. It should be evaluated against conventional alternatives, including the solvers and heuristics already embedded in enterprise products or custom code. Performance can depend heavily on how a business problem is formulated, the quality of input data, the required accuracy, time limits, and integration overhead.
For Windows-focused IT teams, the sensible question is not whether an offering carries a quantum label. It is whether it improves a specific workflow enough to justify adoption. A useful pilot should define a baseline, such as current planning time, cost, capacity utilization, or route quality; use representative data; specify acceptable solution quality; and account for the operational work of moving data between Windows applications, databases, cloud services, and the optimization environment. Security, data residency, licensing, support commitments, and repeatability matter as much as benchmark claims.
This approach also protects buyers from a common category error: treating quantum-inspired software, quantum annealing hardware, and universal gate-based quantum computing as interchangeable. They are related fields, but their capabilities, maturity, deployment requirements, and risks differ substantially.
The policy and enterprise implications
If the reported NEC withdrawal is confirmed in fuller detail, it would underscore how difficult it is to sustain national hardware ambitions through private-sector investment alone. Building machines demands patient capital and a long horizon; producing economic value through software, services, and hybrid systems can have a shorter path to customers. Policymakers considering quantum strategy will need to balance support for frontier hardware research with support for workforce development, applications, classical high-performance computing, and open access to multiple hardware types.
For enterprises, the news should not be read as a reason to abandon quantum-adjacent experimentation. It is a reason to tighten vendor due diligence. Ask whether a supplier is providing hardware, access to partner hardware, classical or quantum-inspired software, consulting, or a hybrid platform. Ask which claims are product commitments rather than research aspirations. And avoid making multi-year architecture decisions based on an assumed hardware roadmap when the public record does not verify it.
NEC’s reported change illustrates a larger market reality: the winners in quantum-related computing may not be determined solely by who manufactures a qubit. They may also be determined by who can turn difficult optimization work into dependable, deployable software on infrastructure organizations can use now. NEC’s continuing Vector Annealing licensing points toward that possibility, while the reported hardware exit highlights the financial and technical barriers that still separate quantum science from a durable hardware business.