Quantum Computing Coexists in Multiple Modalities, Cloud Model Dominates, Tech Giants Hold Significant Advantage

Overview

This report is a summary of a quantum computing industry expert webinar hosted by Bernstein. The expert has 15 years of experience in quantum computing and large technology companies, providing in-depth analysis of current quantum computing technology routes, application prospects, competitive landscape, and timeline. The core conclusion is that the quantum computing industry will not rapidly converge on a single architecture as classical computing did, but will evolve into a multimodal coexisting ecosystem. Superconducting, trapped ion, and neutral atom are currently the three most competitive technology routes, each with advantages in different application scenarios. Although large technology companies have established certain barriers in software and ecosystem, competition in hardware quality remains intense. Additionally, due to extremely high capital barriers, cloud services will become the primary delivery model.

Core views

Technology Route and Advantage/Disadvantage Analysis: Qubits are mainly divided into 'Manufactured Qubits' and 'Natural Qubits'. Manufactured qubits (e.g., superconducting, silicon spin, topological): Built from engineered systems, they operate extremely fast (kilohertz to megahertz levels), suitable for computing tasks requiring large amounts of repeated sampling. However, they face high noise, short coherence times, and manufacturing consistency challenges (e.g., superconducting qubit manufacturing variation is approximately 10%). Currently, superconducting technology leads in scale, speed, and industrial maturity, receiving strong support from large technology companies. Natural qubits (e.g., trapped ion, neutral atom): Utilize inherent physical properties of nature, offering high consistency, long coherence times, and strong noise resistance. Their main drawback is slow operation speed (maximum repetition rate of approximately 1Hz), making high-frequency sampling difficult. However, they have natural advantages in long-range interactions and multi-qubit connectivity, facilitating more efficient error correction codes. Other routes: Photonic quantum computing and silicon spin remain in early stages, facing interaction and control challenges; topological qubits (e.g., Majorana fermions) theoretically possess natural error protection capabilities, but even single-qubit demonstration has not yet been achieved, representing a high-risk, high-reward long-term bet. Application Scenarios and Technology Matching: The trade-off between speed and precision determines application suitability. Applications requiring estimated real-number outputs and relying on large amounts of repeated sampling (e.g., chemical simulation, materials science, drug discovery): Better suited for fast superconducting systems. Even if single-shot precision is not high, results can be approximated through large amounts of sampling. Applications requiring discrete correct answers and extremely high precision (e.g., cryptography breaking, combinatorial optimization): Better suited for high-fidelity natural qubits (trapped ion, neutral atom). These applications typically require running only once to obtain the correct result, making them insensitive to speed but extremely sensitive to error rates. Competitive Landscape and Business Model: Capital-intensive and cloud-dominated: Building and maintaining large-scale quantum computers is extremely costly (e.g., electricity costs alone may reach $100,000 per operation), making widespread hardware ownership unrealistic. The primary business model will be cloud-based access (e.g., AWS Braket, Azure Quantum, IBM Cloud). Advantages of large technology companies: IBM and Google, among other large technology companies, hold favorable positions in competition due to their strong capital, existing infrastructure, and ability to attract top talent. Particularly IBM, whose Qiskit software framework and cloud ecosystem constitute extremely high entry barriers, making it difficult for competitors to surpass in terms of usability and developer community. Paths for startups: It is difficult for startup companies to compete head-to-head with giants at full scale. Their paths to success may include: specialization in specific technology areas, partnerships with large companies, or acquisition. Additionally, some countries (e.g., France, India) support domestic quantum companies through sovereign investment, which also provides survival space for startups. Timeline and Milestones: Short-term (next 5 years): Limited quantum advantages are expected in specific scientific research areas, particularly in hybrid quantum-classical computing, where quantum processors serve as part of high-performance computing (HPC) systems for data preprocessing or feature extraction, rather than fully replacing classical computing. Long-term (15-20 years): There remains a long way to go to achieve large-scale fault-tolerant quantum computers capable of running complex tasks such as Shor's algorithm. NIST estimates related threats may emerge around 2035, while experts believe it may take longer. Key indicators to watch: Investors should focus on demonstrations of logical qubits, implementation of logical gates, progress in scalable error correction systems, and miniaturization of control electronics (e.g., CryoCMOS) as technical milestones, rather than merely focusing on the number of physical qubits.

Analysis framework

The institution uses a qualitative analysis approach through direct dialogue with senior industry experts to梳理 the complex landscape of the quantum computing industry. The analytical approach is as follows: 1. **Technology Classification Framework**: Starting from the physical foundation, qubits are divided into 'manufactured' and 'natural' types, explaining the fundamental differences between different technology routes in core indicators such as speed, noise, and coherence time. 2. **Application Scenario Mapping**: Combining technical indicators with actual application needs, distinguishing between 'high-frequency sampling' and 'high-precision single-shot result' applications, thereby deriving potential winners for different technology routes. 3. **Business and Economic Analysis**: Considering the extremely high capital expenditure (CAPEX) and operating expenditure (OPEX) of quantum computing, deducing that cloud service models will inevitably become mainstream, and analyzing the structural advantages of large technology companies over startups. 4. **Milestone Tracking Method**: Abandoning vague predictions of specific commercial application落地, instead proposing a series of specific, quantifiable technical milestones (e.g., logical qubits, logical gates, control chip integration) as benchmarks for judging industry progress.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply-demand framework

  Divergence in quantum computing supply-side technology routes and matching with demand-side application scenarios

  The report analyzes which technology routes are more competitive in specific areas by distinguishing different types of qubits (supply) and the types of computing tasks they are suited for (demand). For example, superconducting is suitable for chemical simulation requiring large amounts of sampling, while trapped ion is suitable for cryptography breaking requiring extremely high precision.

• Competition and Strategy FrameworkMoat / competitive advantage

  Entry barriers formed by software ecosystem and first-mover advantage

  The report emphasizes that IBM's ecosystem built through Qiskit software and early cloud platforms is its core moat. This first-mover advantage and developer habits make it difficult for later entrants to shake its market position even if their hardware performance is superior.

• Cycle and Prosperity FrameworkInflection Point Analysis

  Technical inflection point from physical qubits to logical qubits

  The report notes that the industry is at a critical stage of transitioning from simply operating physical qubits to building fault-tolerant logical qubits. The degree of realization of this technical inflection point (e.g., demonstration of logical gates) is an important signal for judging whether the industry is moving toward mature commercial applications.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• IBM

  Beneficiary: Software Ecosystem Leader

  Strengths

  Owns Qiskit, the de facto standard programming framework, with early cloud platform deployment, establishing a strong developer community and talent pool.

  Weaknesses

  Hardware quality is being challenged by trapped ion and neutral atom technologies in certain metrics, no longer the absolute hardware performance leader.

  Comparison

  Compared to pure hardware startups, IBM has stronger financial resources and full-stack integration capabilities; compared to Google, its open-source software ecosystem layout is more profound.

  Risks

  If the hardware performance gap becomes too large, users may switch to competitor platforms offering higher fidelity computing.

• Google

  Beneficiary: Strong Technical Capabilities

  Strengths

  Possesses deep technical accumulation and a large R&D team (approximately 1,000 people) in the superconducting quantum computing field.

  Weaknesses

  Started later in software usability and developer ecosystem, currently working to catch up.

  Comparison

  Similar to IBM in terms of strong capital and talent advantages, but slightly inferior in software interfaces and user-friendliness.

  Risks

  Faces dual competition from other superconducting camps and emerging technology routes.

• Rigetti / IonQ / QuEra and other startups

  Divergence: Specialization or Acquisition Targets

  Strengths

  May possess unique technical advantages or higher hardware performance in specific technology routes (e.g., superconducting, trapped ion, neutral atom).

  Weaknesses

  Lack the capital scale and infrastructure of large technology companies, making it difficult to independently bear the R&D costs of large-scale fault-tolerant computing; face talent recruitment difficulties.

  Comparison

  More flexible than giants but weaker risk resistance. Some companies (e.g., QuEra) perform prominently in specific areas (neutral atom).

  Risks

  Funding chain rupture risk; if unable to demonstrate the long-term scalability of their technology route, may be marginalized or acquired.

Key data

• Superconducting Qubit Operation SpeedKilohertz to Megahertz (kHz-MHz)Several orders of magnitude faster than natural qubits, suitable for high-frequency sampling tasks.
• Natural Qubit Operation SpeedMaximum approximately 1HzLimited by physical characteristics, slow repetition rate for repeated computing, but high fidelity.
• Superconducting Qubit Manufacturing VariationApproximately 10%Reflects the consistency and noise control challenges of manufactured qubits.
• Expected Time for Large-Scale Fault-Tolerant Quantum Computing15-20 yearsExpert estimate of the time span required to achieve fault-tolerant computers capable of running complex tasks such as Shor's algorithm.
• Estimated Electricity Cost per Cryptography Breaking Operation$100,000Highlights the extremely high operating costs of quantum computing, supporting the necessity of cloud service models.

Impact & implications

For investors, the quantum computing industry remains in early stages, with technology routes not yet finalized and relatively high investment risk. In the near term, large technology companies (e.g., IBM) are more likely to capture value in the early commercialization phase due to their software ecosystems and cloud platform advantages. Startups need to achieve breakthroughs in specific technology areas (e.g., neutral atom, trapped ion) or exit through acquisition. As the industry moves toward fault-tolerant computing, companies that can provide key supporting technologies (e.g., cryogenic control chips, laser systems) may also present opportunities. Additionally, the maturity of software and application development tools will be a bottleneck constraining industry development, and progress in related areas is worth close attention.

Risks

• Technology route uncertainty: No single technology route has been proven to ultimately prevail; betting on the wrong route may lead to losses.
• Timeline delays: The time to achieve large-scale fault-tolerant quantum computing may far exceed expectations, leading to slow commercialization progress.
• Software bottleneck: The maturity of application programming interfaces and development tools lags behind hardware development, limiting the落地 of actual applications.
• Capital consumption: High R&D and operating costs may cause some startups to fail to survive.

What to watch

• Demonstration of logical qubits and their fidelity.
• Implementation of logical gate operations, particularly protected operations between different logical qubits.
• Progress in scalable error correction systems.
• Breakthroughs in miniaturization of control electronics (e.g., CryoCMOS).
• Improvements in neutral atom and trapped ion technologies in long-range interactions and fast sampling.