Which quantum computing architecture is closest to commercial scale?

No architecture can be ranked through one headline metric. Superconducting circuits offer fast gates and benefit from semiconductor-style fabrication experience, but require deep cryogenics and complex wiring. Trapped ions can deliver high-fidelity operations and flexible connectivity, while optical control and gate speed create different scaling challenges. Neutral atoms offer large configurable arrays, photonics offers manufacturing and networking possibilities, and spin-based approaches seek compatibility with semiconductor processes. Each platform exchanges advantages in physics for constraints in engineering.

The relevant unit of comparison is a useful workload executed under disclosed conditions. Physical qubit count, logical qubit count, circuit depth, gate fidelity, connectivity and runtime describe different properties. Error mitigation can improve near-term results without creating fault tolerance. Error correction introduces substantial physical and classical overhead, and its value depends on demonstrating that encoded performance improves as code distance or system scale increases. Claims should therefore identify the benchmark, comparator, error model and total resources consumed.

Commercial scale is also workload-specific. Annealing, analogue simulation, sensing and universal gate-model computing address different problems and should not be placed on a single maturity ladder. A system can create value for a narrow optimisation or simulation task without establishing general-purpose advantage. Conversely, a broad fault-tolerant roadmap may have greater theoretical reach but a longer and more capital-intensive path to customers.

Vendor roadmaps remain useful when treated as primary-source statements of intent. IBM's published roadmap, for example, makes explicit claims about modularity, logical operations and fault-tolerant milestones, while trapped-ion, neutral-atom and photonic developers publish different scaling hypotheses. Independent programmes such as DARPA QBI are important because they evaluate whether the complete engineering plan can produce utility relative to cost. QFM separates company claims, peer-reviewed evidence and independent validation rather than collapsing them into one confidence level.

Architecture choice determines organisational design as well as physics. A vertically integrated company may control fabrication, packaging, control and software to optimise the whole system, but it must finance every layer. A modular strategy can use specialist suppliers and iterate faster, while creating interface dependencies and weaker control over scarce capacity. Investors should identify which layers are proprietary, which are purchased, and whether the scaling roadmap depends on suppliers whose own capacity or specifications remain unproven.

A credible comparison therefore uses an evidence hierarchy. Peer-reviewed results with disclosed methods rank differently from company demonstrations; independent customer workloads rank differently from simulated projections; and complete-system performance ranks differently from one component metric. The hierarchy does not make company data irrelevant. It makes the provenance and scope of every claim explicit, allowing new evidence to update the assessment without turning each announcement into a categorical verdict.