Competition authorities are gearing up to regulate quantum computing. The problem: there is no market there yet.
In March 2026, the Italian Competition Authority (AGCM) launched an “IC59 fact-finding inquiry” into quantum, citing concerns that ongoing developments could shape long-term competition. Drawing on lessons from artificial intelligence (AI) and cloud markets, the AGCM flagged familiar risks: lock-in, technological preemption, and barriers to knowledge and entry.
The move reflects a broader shift in European competition policy following the Digital Markets Act (DMA). Regulators now focus less on prices and market share, and more on how markets are designed and governed—especially where control over key inputs may entrench gatekeepers and confer durable advantages.
That shift carries real risks for emerging technologies like quantum computing. The AGCM casts its inquiry as a “timely reconnaissance” of a nascent market—exploratory, not enforcement-driven. Even so, treating the quantum ecosystem as if it were already a mature market risks overstating what we can know about its trajectory. The technology remains too uncertain for reliable market analysis. Any risk assessment necessarily rests, at least in part, on incomplete and uncertain information, raising the prospect of unintended consequences for technological development.
Competition law rests on familiar assumptions: markets generate observable signals, and analysts can use those signals to assess competitive dynamics and identify harm. Authorities evaluate conduct—pricing, access restrictions, exclusionary agreements—against established benchmarks grounded in current market conditions and evidence. In quantum computing, however, market structures remain too underdeveloped to support meaningful competition analysis.
Premature Market Definition Is the Root of Many Errors
What exists today in quantum computing is not a “market” in any meaningful competition-law sense. It is a hybrid ecosystem of large technology firms, startups, and publicly funded research programs—shaped less by competitive equilibrium than by financing patterns, experimentation, and deep technological uncertainty.
When signals of efficiency, substitution, and consumer preference remain incomplete or unstable, the inferences drawn from them become fragile. That fragility increases the risk of mistaking technological development for anticompetitive conduct. Authorities may read experimentation as exclusion, partnerships as foreclosure, and vertical integration—often necessary in hardware-intensive technologies—as evidence of dominance. Apply competition law too early, and you risk injecting regulatory uncertainty that chills the very investment needed to move the technology forward.
The AGCM’s inquiry hints at a deeper problem: an analytical framework that does not quite fit. That mismatch risks producing flawed regulation, confused enforcement, and slower market development. Competition authorities often face the temptation to treat emerging technologies as if they already constitute a single, unified market. Quantum computing does not. It is far from clear that it ever will.
A better way to understand quantum computing is as a stack of loosely connected layers: hardware, software, algorithms, and domain-specific applications. Each layer evolves at its own pace and faces distinct constraints—from physics and engineering to mathematics and economic viability. Progress is uneven and interdependent. Advances in one layer do not automatically unlock gains in others, which helps explain the slow pace of commercialization. In that sense, quantum computing functions more like enabling infrastructure than a discrete market. It does not fit neatly into the tidy categories regulators prefer.
The same ambiguity appears in AI. Generative AI, in particular, is not a single, unified technology, but a loose collection of models, techniques, and systems built on an evolving stack. These components operate as inputs in broader pipelines, supporting a wide range of downstream applications, business models, and user-facing functions. What matters is not the structure alone, but how these systems are integrated, orchestrated, and deployed..
Both quantum computing and generative AI are better understood as assemblages of loosely connected technological layers, not monolithic systems. Their differences lie in use and combination, not in the existence of a stack. Over time, quantum computing will likely integrate into existing computational pipelines, potentially alongside AI systems that orchestrate workflows and decision-making. That is not a novel development—it is another instance of technological stacks interacting and co-evolving. In the near term, neither technology is likely to consolidate into a single, unified system.
Quantum technologies also share underlying scientific principles while serving distinct applications. Innovation in one domain will likely spill over into others. Approaches that seem suboptimal in one context may prove highly valuable in another. That reality undermines the idea of a single, well-defined future “quantum market” and risks obscuring important cross-domain complementarities—both within quantum technologies and across adjacent fields. Early intervention, then, risks foreclosing design choices that could ultimately prove welfare-maximizing.
Seeing Monopoly in a Pre-Market
A limited understanding of how the quantum-computing stack evolves can distort assessments of market concentration and power. More troubling, those assessments often get projected onto a market that does not yet meaningfully exist.
The Italian Competition Authority (AGCM) already flags the concentration of activity among a small number of well-resourced actors and draws an analogy to artificial intelligence (AI). That comparison does not hold. AI is already deployed commercially—albeit not at its most advanced stages—while quantum computing remains largely pre-commercial and research-intensive. What looks like concentration instead reflects technological constraints: high costs, uncertain development paths, and limited commercial returns. The sector is expanding, but from a very small base. Opening formal inquiries at this stage risks getting ahead of both the technology and its market implications.
Why Concentration Misleads
Even in AI, concentration is more complicated than it first appears. High development costs favor larger firms, but regulation reinforces that dynamic. As compliance costs rise, the ability to meet regulatory requirements increasingly determines which firms can scale. The result is a market structure with a concentrated core and a broader layer of smaller, innovative firms—often linked back to the core through partnerships, licensing, or acquisitions.
Seen in that light, concentration alone is a weak signal of market power. In quantum computing, it is weaker still. Apparent concentration reflects technical complexity, specialized expertise, collaboration, and high-risk investment—not clear evidence of anticompetitive conduct. Commercial maturity and technological capability are not moving in tandem. That does not eliminate the possibility of future competition concerns, but it does weaken the case for early intervention. On this basis, the AGCM’s concerns about barriers to entry may be overstated.
Competition in this setting turns less on firm count than on whether the ecosystem sustains continuous innovation. Treat concentration as a problem, and you risk misreading how competition actually works in an emerging field. Capital clusters around a small number of players not because they control a market, but because they can absorb the costs and uncertainty of long-term development.
Architecture Matters: Interoperability Over Control
The structure of the quantum ecosystem further weakens any inference that concentration will translate into durable market power. Quantum technologies are not developing as vertically integrated silos. Instead, they rely on interoperable, hybrid architectures that combine classical and quantum resources.
Initiatives such as the Munich Quantum Software Stack and platforms like NVIDIA’s CUDA-Q reflect a deliberate move toward hardware abstraction, allowing developers to write code that runs across multiple quantum back ends—from simulators to physical processors. Open-source tools—including the Guppy programming language and the Selene emulator released by Quantinuum—push in the same direction. Where access to hardware remains limited, innovation shifts to the software layer by widening participation.
This separation between software and hardware lowers switching costs and limits the extent to which control over a specific quantum processor can create user lock-in. Even where hardware capabilities remain concentrated, access increasingly runs through cloud-based interfaces and standardized development environments. Cloud platforms act less like gatekeepers and more like integrators of heterogeneous back ends, allowing users to interact with multiple quantum technologies through a single interface.
The quantum cloud does not map neatly onto traditional antitrust frameworks. Several major providers—notably Amazon Web Services (AWS) and Microsoft—act as aggregators, competing to offer broad access to diverse quantum processors. Others, such as IBM and Google, remain more vertically integrated. Platforms like Scaleway’s QaaS unify multiple European quantum modalities under a single cloud-native interface, allowing developers to switch between GPU clusters and quantum hardware without changing their development environment. Vendor-independent middleware, including Q-AIM and Qunicorn, enables translation across different quantum-circuit and result formats, reducing friction when running workloads across competing clouds. The value of the quantum cloud lies in integration, not exclusion.
No Gatekeepers (Yet)
As platforms aggregate more processors, they become more useful—not because switching is costly, but because they reduce the complexity of accessing multiple systems. That dynamic can be pro-competitive. Platform power is constrained by users’ ability to multi-home, by rival platforms offering similar integrations, and by open-source middleware that continues to lower switching costs.
These dynamics extend to system architecture. Research efforts, including middleware developed at the KTH Royal Institute of Technology, aim to decouple applications from underlying hardware, allowing programs to run across platforms with minimal adjustment. IBM’s quantum architecture incorporates vendor-agnostic layers that abstract hardware-specific features, while platforms such as Quantum Rings provide cloud-based access to multiple processors.
Taken together, these developments point to an ecosystem in which interoperability is not incidental, but engineered. By lowering switching costs and enabling multi-homing, they limit any single firm’s ability to convert control over hardware or infrastructure into market power. While still evolving, these dynamics weaken the link between concentration and control that conventional indicators assume. The modular structure reduces the scope for any one firm to internalize the entire value chain and instead promotes interdependence across layers. Emerging open-source approaches reinforce this trend by lowering coordination costs, facilitating benchmarking, and expanding the pool of developers—further complicating efforts to define clear market boundaries.
The AGCM itself acknowledges this structure, noting that the sector “features both global big tech companies providing cloud-based services and small to medium-sized players, often still start-ups, focused on developing specialised technologies and services.” This division of labor is not incidental; it is structurally significant. If control over quantum hardware were enough to secure dominance, one would expect tighter vertical integration and the marginalization of independent actors. The continued presence of specialized firms across multiple layers instead suggests that control remains distributed and that complementarities—not consolidation—are shaping the sector’s development.
This also cuts against concerns about gatekeeping, often framed through a DMA-style logic. For example, Pasqal’s neutral-atom quantum processing units (QPUs) are already accessible across five major cloud platforms, including Azure Quantum, Google Marketplace, OVHcloud, and Scaleway. This kind of multi-platform distribution places quantum hardware “where users already work,” weakening the idea that access will be tightly controlled by a single provider.
At the same time, emerging approaches may further reduce the need for direct hardware ownership. Equal1, for instance, aims to develop quantum processors using existing semiconductor fabrication infrastructure, bringing semiconductor-scale manufacturing to quantum computing. This model challenges the assumption that access requires bespoke, vertically integrated infrastructure controlled by a few dominant players. Instead, it distributes capabilities across different layers of the stack. The result is a fluid and uncertain ecosystem—one that weakens claims that current patterns of concentration will translate into durable market power.
Jumping the Gun on Quantum
None of this dismisses the Italian AGCM’s underlying concern. Early design choices can shape long-term market outcomes. The question is not whether to intervene early, but whether competition-law tools fit a setting where their core assumptions—clear boundaries, stable roles, and identifiable dominance—have yet to take hold.
Intervene too soon, and you risk forcing an ill-suited framework onto a still-forming ecosystem. That framework tends to privilege market structure—counting firms, measuring concentration, and inferring that structure determines conduct, and in turn, performance. But market structure is itself an output of the competitive process. In quantum computing, that process remains in flux across multiple sectors.
A more fitting response may lie in anticipatory governance, rather than extending competition-law reasoning into terrain where its assumptions do not yet hold. Given quantum computing’s enabling nature, concerns around access, control, and conduct may already fall—at least in part—within existing sector-specific frameworks. They do not automatically warrant a competition-law response.
Competition law works best ex post: when markets have formed, conduct is observable, and harm can be demonstrated. Apply it too early, and speculation substitutes for evidence. Worse, you risk slowing markets before they have a chance to emerge.
