Is Japan Investing Too Early in Quantum? Lessons from AI History on Strategy and Sustained Funding
Japan is investing in quantum technologies at one of the largest scales globally.
At the same time, quantum computers remain under development, and practical computational capabilities that definitively outperform conventional computers have not been widely demonstrated.
Given this high level of uncertainty, is it necessary to pour large amounts of public money into the field now? Wouldn’t it be better to invest once the technology matures?
Japan’s history in AI research offers instructive insights for this question. Reviewing current quantum policy alongside that history reveals the rationale for early investment and highlights the importance of sustaining funding over the long term without depending on excessive expectations.
How much is Japan investing in quantum technologies?
The Japanese government formulated the “Quantum Technology Innovation Strategy” in 2020 and designated quantum technologies as a national priority area.
The 2022 Quantum Future Society Vision set goals for 2030 including 10 million domestic users of quantum technologies, roughly ¥50 trillion in production value attributable to quantum technologies, and the creation of unicorns in the quantum field. The ¥50 trillion figure is not meant as quantum-computer sales revenue but as a broad estimate of production value across industries to which quantum technologies may contribute.
The scale of investment is also substantial. In its 2024 supplementary budget, the Ministry of Economy, Trade and Industry (METI) announced quantum-computer industrialization support measures totaling ¥100.9 billion, including government-guaranteed obligations. A further approximately ¥100.4 billion was allocated in the 2025 supplementary budget. These figures do not represent single-year cash outlays.
Other countries such as the United States, China, the United Kingdom, Germany, and France are also making large investments, so Japan is not unique in this respect. However, simple comparisons are difficult because each nation’s budgets differ in scope and accounting periods.
Meanwhile, commercialization of quantum computers remains at an early stage. Although research on quantum error correctionquantum error correction量子誤り訂正 / Quantum Error Correction / QECA technique that distributes information across multiple physical qubits and detects and corrects errors without directly disturbing the quantum state.QI NoteSimply implementing it does not automatically provide practical fault tolerance. What matters is whether the logical error rate is improved relative to the physical error rate. and logical qubitslogical qubitsLogical Qubit / Logical QubitA unit of information treated as a single qubit protected from errors by using multiple physical qubits and quantum error correction.QI NoteSimply having “created a logical qubit” does not necessarily mean fault-tolerant quantum computing (FTQC) has been achieved. One should verify logical error rates, operational/gate performance, and scalability. has progressed, broadly available practical computation services that surpass classical computers are not yet established.
Judged only by current economic outcomes, it is hard to justify such large investments. But the value of R&D investment cannot be measured solely by short-term returns.
Japan once led the world in AI research
Japanese researchers made significant contributions to the neural-network research that underpins today’s AI.
Shun’ichi Amari studied learning theory for neural networks in the 1960s and in 1967 formalized learning principles using gradients. Kunihiko Fukushima’s 1980 paper on the “Neocognitron” is known as a pioneering model that anticipated today’s convolutional neural networks (CNNs).
Japan was by no means a latecomer in AI research.
However, Japan fell behind countries such as the United States in forming the large companies and computing infrastructure that would later drive the AI industry.
Producing pioneering research and nurturing it into an industry require different capabilities.
There is another notable national project in Japan’s AI history.
It is the Fifth Generation Computer Project, which began in 1982.
At the time, the Ministry of International Trade and Industry invested about ¥54 billion over roughly 11 years to develop new computers capable of knowledge processing and logical reasoning.
The research produced results such as parallel inference machines and logic-programming techniques. However, subsequent AI development centered on statistical machine learning and neural networks.
The 1993 final evaluation report assessed the academic and technical achievements, while a 1993 research evaluation noted that the project failed to produce sufficiently useful applications for end users and that the adopted technology stack did not become mainstream.
It would be inappropriate to label the Fifth Generation project simply a failure: it left legacies in basic research and human-resource development.
However, there is an important lesson here.
Even if a nation invests over a long period and achieves its R&D goals, that technology may not become the mainstream in future industries.
This issue overlaps with the current quantum-computer development landscape, where multiple technical approaches are competing.
Does Japan’s quantum policy learn from past lessons?
Is the government concentrating massive funds on a single quantum-computer approach?
In practice, it is not.
Under METI and NEDO’s post-5G-related programs, system development for multiple approaches—superconducting, neutral-atom, and photonic systems—is being supported. In 2026, companies developing ion-trap systems such as Qubitcore and QUDORA Japan were also selected for R&D on related component technologies.
| Technology approach | Major domestic awardees/examples | Examples of supported areas |
|---|---|---|
| Superconducting | Fujitsu | Quantum computer system development |
| Neutral atoms | Yaqumo, QuEra | Quantum computer system development |
| Photonic | OptQC | Photonic quantum computerPhotonic quantum computer光量子コンピューター / Photonic Quantum Computer / Optical Quantum ComputerA quantum computer that performs quantum computation using photons or the quantum states of light as carriers of information.QI NotePhoton generation, loss, detection, and integration affect performance. One should consider not only the number of qubits but also loss rates and the performance of light sources and detectors. development |
| Ion-trap–related | Qubitcore, QUDORA Japan | Element technologies such as optical interconnects and ultra-low-temperature packaging |
*The listed awardees are representative examples within the post-5G program and do not form an exhaustive list. Budget sizes and development targets differ across programs.
Moreover, funding is being allocated not only to quantum computers themselves but also to ultra-low-temperature technology, optical devices, control equipment, middleware, and other related areas.
Notably, this investment approach is not accidental.
METI’s June 2026 interim evaluation materials explicitly state that which technical approach will be realized in the future is uncertain, and that concentrating on a single approach carries risk.
At the same time, the materials explain why METI prioritized system-development support for three approaches—superconducting, photonic, and neutral-atom—based on the maturity of domestic actors. Support for ion-trap and other related technologies is proceeding under separate development frameworks.
In other words, the Japanese government is not simply distributing funds evenly across all approaches. It sets priority areas based on current technical strengths while maintaining multiple options in parallel.
Furthermore, the same materials mention the possibility of narrowing support targets as technological progress clarifies the landscape.
This is an important difference from the Fifth Generation Computer project. Unlike past national projects that centered on a specific computer architecture and technology stack, current quantum policy is designed on the premise that the winning technical approach is uncertain.
Of course, whether actual budget allocations are optimal is another question. But at least in its current policy design, Japan avoids betting on a single approach before a winner is clear and instead pursues a strategy of developing multiple approaches and related industries in parallel.
More important than early investment is whether funding can be sustained
The idea that one can wait to start investing until quantum computers are practical also has problems.
Quantum-computer development requires not only quantum physics and information science but also ultra-low-temperature engineering, semiconductor manufacturing, precision metrology, electronic circuits, and control software, among other diverse technologies.
Training researchers and building research infrastructure takes many years. If intellectual property, manufacturing know-how, and industrial networks accumulate in early-adopter countries, it may be difficult to catch up even with later financial investment.
Moreover, technologies developed along the way can be applied beyond quantum computers. Early investment thus has solid justification.
However, starting investment is not enough on its own.
AI research history shows cycles in which high expectations lead to disappointment about practical limits and a subsequent contraction of funding and public interest—the so-called “AI winters.” Neural-network research that continued through periods of low attention eventually provided the foundation for later AI progress.
Quantum computing requires a similar long-term perspective.
In fact, this issue is already the subject of discussion in Japanese policy evaluation.
METI’s June 2025 administrative project review called for concretizing roadmaps to achieve the 2030 outcome targets and also recommended considering measures beyond 2030.
The review also acknowledged that ambitious targets should be revisited based on research progress, and that failure to meet specific targets does not necessarily render the entire program meaningless.
Crucially, this separates the policy milestone of 2030 from the technical maturity of quantum computers.
R&D progress does not necessarily follow policy deadlines. Even if practical quantum computers are not fully established by 2030, the accumulated technologies, human resources, and industrial base up to that point do not become valueless.
Conversely, domestic manufacturing of quantum computers alone does not guarantee industrial success.
Therefore, sustained investment does not mean funneling unconditional funding to the same research topics or companies indefinitely.
METI’s R&D evaluation framework includes interim evaluations and mechanisms for plan revisions, anticipating support adjustments based on technological and market trends.
In institutional design, Japanese quantum policy aims to balance long-term investment with flexible technology choices. What will be tested going forward is whether these mechanisms can be effectively applied in policy implementation.
Also important is guarding against excessive short-term hype that suggests quantum computers will accelerate all computations in the near future. If short-term promises are overblown and fail to materialize, funding will be cut. Such a cycle would negate the advantages of early investment.
What matters for Japan’s quantum investment is not waiting until the technology fully matures, nor rigidly sticking to initial development plans. It is accepting uncertainty and continuing to invest in R&D and industrial foundations while adjusting funding targets as technologies evolve.
NETWORKSee the industry network surrounding quantum computing technologies →
