Quantum computing has reached its "fab moment". Who will benefit first?
Recent intensive actions in quantum computing are no longer just scientific research news. The U.S. Department of Commerce announced through NIST that it plans to provide a total of $2.013 billion in federal incentives to 9 companies under the CHIPS and Science Act to advance the key technologies required for practical-scale, fault-tolerant quantum computing. Among them, IBM is expected to receive $1 billion to establish an independent quantum wafer foundry, Anderon; GlobalFoundries is expected to receive $375 million to build domestic quantum foundry capabilities in the United States. Almost simultaneously, imec demonstrated what it claims to be the world's first quantum dot qubit device fabricated using High NA EUV lithography and emphasized that the device is compatible with 300mm wafer fabs.
The capital market is also sending signals. Quantinuum updated its IPO filing, planning to issue approximately 21.05 million Class A common shares with an offering price range of $45 to $50, and plans to list on the NASDAQ under the ticker symbol "QNT". Calculated at the upper end of the offering price, the fundraising scale is approximately $1.05 billion. It is estimated that based on the post-issuance share capital and the upper limit of $50, Quantinuum's listing valuation could reach up to approximately $12.7 billion. The same report revealed that the company had revenues of $30.93 million and a net loss of $458.2 million in 2025, and revenues of $5.24 million and a net loss of $128.2 million in the first quarter of 2026.
Putting all this information together, it's not just about "another breakthrough in quantum computing". What really deserves attention is that quantum computing is moving from scientific research prototypes, cloud demonstrations, and algorithm concepts to a stage closer to the essence of the semiconductor industry. In the future, policy funds and industrial capital will increasingly be invested in wafer fabs, PDK, MPW, online testing, low-temperature CMOS, silicon photonics interconnection, advanced packaging, material interfaces, and system integration.
What is the United States betting on?
The most important information in this round of U.S. quantum incentives is not just the amount of money, but the flow of funds. The two largest funds are being directed towards IBM/Anderon and GlobalFoundries, which means that policymakers have regarded quantum computing as an advanced manufacturing supply chain that needs to be built in advance. IBM/Anderon corresponds to a 300mm quantum wafer foundry, mainly serving the manufacturing of superconducting qubits and related electronic wafers; GlobalFoundries is positioned as a secure domestic quantum foundry, covering multiple architectures such as superconducting, ion trap, photon, topological, and silicon spin.
The support directions for the remaining companies are also clearly more engineering-oriented, rather than just focused on algorithms or software. Atom Computing and Infleqtion focus on the system integration of neutral atoms, Diraq focuses on the integration of silicon spin quantum logic units and manufacturing, PsiQuantum is involved in electro-optic materials, single-photon detectors, and low-loss photon packaging, Quantinuum focuses on integrated photonics and reliable optical components in the ion trap route, and Rigetti is involved in the miniaturization of readout electronics and the architecture of next-generation low-temperature systems. D-Wave stated in an official announcement that the proposed funds will advance its superconducting annealing and gate model systems and explicitly mentioned that expanding quantum computing systems requires advanced manufacturing and packaging technologies. PsiQuantum emphasized that the funds will be used to improve the domestic manufacturability and performance of key components such as BTO high-performance optical switches, high-temperature single-photon detectors, and advanced packaging.
This indicates that the first thing that quantum computing commercialization may bring is not "large-scale shipments of quantum CPUs", but a re-evaluation of the demand for low-temperature control chips, special interconnections, low-loss materials, packaging substrates, radio frequency/microwave readout, silicon photonics components, and superconducting processes. IBM says that Anderon will provide capabilities such as superconducting wiring, through-silicon vias, bumps, dedicated PDK, online wafer testing and characterization, and benchmark process routes. These terms belong to the language of the mature semiconductor industry, not the language of the laboratory. For quantum hardware to truly expand, it cannot rely on a small number of laboratory samples and manual debugging for a long time. Instead, it needs to establish reusable processes, design rules, batch metrology, failure analysis, and yield learning curves similar to the CMOS industry.
GlobalFoundries' statements also point in the same direction. The company said that its Quantum Technology Solutions will manufacture complete quantum hardware solutions from quantum processor units to low-temperature readout and control ICs, and then to advanced packaging and superconducting interconnections, and emphasized low-temperature CMOS, FDX platforms, materials science, and multiple qubit routes. This means that traditional wafer fabs are trying to transform quantum computing requirements into a manufacturable platform that can be serviced, charged for, and iterated. For the semiconductor industry, this is more important than a single-point experiment because only when a manufacturing platform emerges can the industrial chain form a stable division of labor.
Quantum chips start to absorb advanced process capabilities
imec's High NA EUV quantum dot qubit device reveals another change in the industrialization of quantum computing: Advanced process capabilities are entering the manufacturability verification of quantum chips. imec said that a practical quantum computer needs to be scaled up to millions of connected qubits; the reason why silicon quantum dot spin qubits are called "industry qubits" is that their manufacturing process is highly compatible with standard silicon CMOS. In this device, the control gate gap is reduced to approximately 6 nanometers, and High NA EUV is used to achieve reliable patterning of gaps at the nanometer scale.
This does not mean that High NA EUV will soon become a necessary condition for the mass production of quantum chips. More accurately, it indicates that the key indicators of qubit devices are entering the scale of advanced semiconductor manufacturing. Line width, spacing, edge roughness, interface defects, parasitic capacitance, process fluctuations, and wafer-level consistency can all directly affect qubit coupling, noise, coherence time, and repeatability. For semiconductor companies, quantum chips are not a simple extension of traditional logic chips, but are absorbing the precision patterning, defect control, metrology feedback, and yield learning capabilities of the mature process system.
The European SPINS test line further reinforces this judgment. Coordinated by imec, the project aims to develop a 300mm Ge/GeSi platform, conduct research-level MPW runs, develop early PDK, perform wafer-level detection, advanced heterogeneous integration, FDSOI-compatible processes, low-temperature standard circuits, and high-throughput low-temperature characterization. This configuration is similar to the path of the semiconductor industry from early R & D to ecological manufacturing: first, there is a test line, then a design kit and multi-project wafers, and then process documentation, test standards, and ecological interfaces are formed. Once quantum computing enters this stage, the competition will not only occur in physical laboratories but also between process platforms and supply chains.
The current quantum hardware routes have not yet converged. Google's Willow continues to use the superconducting route, emphasizing a decrease in error rates as the array scales up, and claims to have achieved progress in quantum error correction below the threshold. Microsoft released Majorana 1, claiming to promote the topological qubit route through a topoconductor material stack composed of indium arsenide and aluminum, and proposed the goal of scaling up to one million qubits on a single chip. In China, a superconducting quantum computer based on the "Zuchongzhi III" design has been made available for commercial use through the "Tianyan" quantum cloud platform, but the claims of relevant quantum advantages or commercial advantages still require more independent verification.
These developments do not indicate that a particular route has won. A more realistic judgment is that quantum computing has entered a stage of "parallel development of multiple technology routes with converging engineering bottlenecks". Although the physical principles of superconducting, silicon spin, photon, ion trap, neutral atom, and topological routes are different, they are all pushing the requirements towards manufacturing, metrology, packaging, low-temperature control, and system integration.
Will the "shovel sellers" in the semiconductor industry benefit first?
From a market perspective, quantum computing cannot be simply compared to AI training chips. The demand for AI accelerators comes from the already commercialized large model training and inference loads, and the customer spending path is relatively clear; Quantum computing is still in a stage of high investment, low income, strong R & D, and strong policy drive. McKinsey estimates that the revenue of quantum computing companies in 2024 is approximately $650 million to $750 million, and is expected to exceed $1 billion in 2025; in 2024, global quantum technology startups raised nearly $2 billion in financing, a nearly 50% increase from approximately $1.3 billion in 2023. These figures indicate a significant warming of the industry, but compared with the AI accelerator, advanced logic foundry, or storage markets, the revenue of quantum computing hardware is still at a relatively small base.
Quantinuum's IPO update makes this contradiction clearer. The company plans to raise up to approximately $1.05 billion, and the capital market reports estimate its valuation ceiling at approximately $12.7 billion. However, the same set of public information also shows that the company's revenue scale is still small and the losses are still large. The report also mentioned that the risk factors for Quantinuum include the need to develop high-volume manufacturing processes and the current dependence on single-source suppliers for several materials and systems. This precisely shows that the valuation of quantum computing companies is not a pricing of current revenues but an early pricing of future fault-tolerant systems, manufacturing expansion, and supply chain integration capabilities.
In the short term, the market is more likely to re-evaluate the "shovel seller" segments first, including low-temperature CMOS, radio frequency/microwave control, silicon photonics, advanced packaging, low-loss materials, special substrates, wafer-level testing, low-temperature probe stations, dilution refrigerators, EDA, PDK services, and HPC interconnections. These segments may obtain new demand from government projects, test line construction, corporate R & D, and small-scale system integration without waiting for the full maturity of general-purpose fault-tolerant quantum computing.
Nvidia's release of NVQLink also reflects this direction. This open architecture is used to tightly couple GPU computing with quantum processors, supports 17 QPU manufacturers, 5 controller manufacturers, and 9 U.S. national laboratories, and emphasizes that quantum error correction and calibration algorithms require low-latency, high-throughput connections to classical supercomputers. This shows that for a long time in the future, quantum computing will not exist independently of the classical computing system. It is more likely to exist as a heterogeneous co-processor in HPC and AI supercomputing centers, with GPUs/CPUs responsible for control, calibration, error correction decoding, simulation, and execution of hybrid algorithms.
The medium- and long-term forecasts are still optimistic. McKinsey predicts that by 2035, the revenue of quantum computing could reach $28 billion to $72 billion, and by 2040, the total market for quantum technology could reach $198 billion. BCG believes that by 2040, quantum computing could create an economic value of $450 billion to $850 billion and support a hardware and software supplier market worth $90 billion to $170 billion. However, BCG also warns that quantum computing has not yet provided tangible advantages over classical computing in commercial or scientific applications, and the insufficient hardware fidelity still limits its adoption, while GPUs, classical algorithms, and AI frameworks are still continuously raising the bar.
Conclusion
Quantum computing is entering a stage closer to the essence of the semiconductor industry. This will bring new opportunities to the equipment, materials, packaging, testing, and foundry segments, but it will not immediately lead to a large-scale release of consumer electronics.
For companies in the industrial chain and investors, quantum computing is worth paying attention to, but it is not suitable to be priced based on the logic of short-term explosion. A more reliable observation path is to look at three things: whether wafer fabs and test lines can continuously produce stable devices; whether quantum PDK, MPW, and online testing can lower the threshold for hardware innovation; and whether low-temperature control, advanced packaging, silicon photonics interconnection, and HPC coupling can form a reusable supply chain capability. Those who can transform physical breakthroughs into wafer fab processes, turn laboratory devices into PDK, and convert single demonstrations into yield curves are closer to the real entry point for the commercialization of quantum computing.
In this sense, the impact of quantum computing on the semiconductor market should not only be judged by long-term application imagination but also by which manufacturing capabilities, material platforms, and system integration segments it is driving to be re-priced. In the short term, it will not become another AI chip market; but in the medium and long term, it may become a new round of technology reserve competition field for advanced manufacturing, low-temperature electronics, silicon photonics, packaging and testing, and HPC system manufacturers. What can truly withstand valuation fluctuations is still manufacturing capabilities, supply chain resilience, and verifiable engineering progress.
This article is from the WeChat official account “Semiconductor Industry Insights” (ID: ICViews), author: Junxi, published by 36Kr with authorization.