An impressive machine, scarce research talent, curious customers: the scene is set. One question remains, less photogenic than the golden cables of a quantum computer: who will pay for the years between a scientific demonstration and a profitable service? Looking ahead to September 2026, the sector’s challenge is as much financial as technical. The documented developments below shed light on this trajectory; the outlook remains a set of scenarios, not established outcomes. For entrepreneurs, the challenge boils down to one principle: funding the wait without turning uncertainty into a marketing pitch.
Quantum technology runs on more than one clock
Talking about the “quantum market” as a homogeneous whole already leads to poor decisions. Computing, communications and sensors share scientific foundations, but not the same customers, industrial constraints or timelines. A magnetometer designed for a specialized application does not have to wait for a universal computer to correct its errors effectively. Conversely, selling optical components does not prove that a quantum computer will quickly become profitable.
In computing, hardware progress is real, but systems still face challenges involving noise, errors and scaling. Google’s announcements on error correction, particularly with Willow in December 2024, illustrated an important advance: under certain experimental conditions, increasing the size of a code can reduce logical errors. This is not yet a demonstration of a competitive industrial service. A scientific milestone can be significant without immediately becoming a product.
This distinction should shape investor presentations. What result has been observed? On what task? Using what classical computing resources as a comparison? How much of the system actually works together? A credible company does not gloss over these questions: it turns them into a work program that can attract funding.
Replace the qubit count with an evidence framework
The number of qubits has become a convenient shorthand. Yet in isolation, it says little about the quality of operations, connectivity, availability or the total cost of a computation. Architectures based on neutral atoms, trapped ions, superconducting circuits or photons involve different trade-offs. Ranking them in a single column is like comparing vehicles solely by their number of wheels.
To fund a roadmap rigorously, it is better to agree on milestones combining physics, engineering and practical use. Fidelity measured under reproducible conditions; a characterized logical operation; a period of stable operation; deployment at a customer site; then a result benchmarked against the best relevant classical method. These stages are not interchangeable, and their criteria must be set before any announcement.
- Technical evidence: document conditions, limitations and repeatability.
- Industrial evidence: measure availability, maintenance and manufacturing capacity.
- Economic evidence: establish what the customer gains, with all costs included.
This evidence framework also protects teams. When objectives are explicit, a delay can be explained and an approach abandoned without rewriting the entire story. A board can then fund a decisive experiment rather than demand an artificially spectacular announcement every quarter.
Early revenues do not all tell the same story
The sector already has ways to generate revenue: cloud access to machines, research contracts, software, integration services, control equipment, lasers and cryogenics. These sales can form a solid business. However, not all of them demonstrate the existence of independent demand for quantum computing. An industrial company paying for an experiment is sometimes primarily acquiring expertise and an option on the future.
It is therefore essential to distinguish exploratory revenue from revenue likely to recur. A subsidized project, a one-off installation and a renewed subscription do not offer the same visibility. Partnerships require similar scrutiny: a logo in a presentation reveals neither the amount committed, nor any purchase obligation, nor technical validation. The useful question remains: would the customer still pay without public support and without a publicity objective?
Sensors and components may offer nearer-term opportunities in some applications, but they too face established alternatives. Additional sensitivity, size, ruggedness and operating costs must be assessed together. A device that excels in the laboratory may lose its advantage if it requires conditions incompatible with a factory, construction site or hospital.
Cybersecurity: a present need, an essential distinction
Preparing for the quantum threat is a separate market. In August 2024, the U.S. NIST published its first three finalized post-quantum cryptography standards. For organizations, the work can begin with an inventory of cryptographic mechanisms, identifying data that needs long-term protection and planning the migration. There is no need to wait for a machine capable of breaking current systems before preparing infrastructure that is difficult to replace.
But post-quantum cryptography runs on classical computers. Its revenues therefore do not automatically validate the business models of quantum computer manufacturers. It also differs from quantum key distribution, which requires specialized equipment. Perpetuating this confusion artificially inflates the commercial narrative; explaining it, by contrast, helps sell the right solution to the right problem.
Patient capital, but not blind capital
Public strategies, including the French plan announced in 2021 and European programs, address a real problem: funding skills, infrastructure and technological sovereignty in which the market alone may underinvest. Public procurement can also provide a testing ground. It becomes particularly useful when it requires comparable assessments and leaves room for multiple approaches.
For a young company, the challenge is to assemble an appropriate funding mix. Grants support scientific uncertainty; equity absorbs industrial risk; customer contracts fund defined deliverables. Debt becomes trickier when revenues remain distant or unpredictable. Financing open-ended research with rigid repayment schedules can turn a normal technical delay into a cash-flow crisis.
Diversification offers breathing room, provided it does not spread teams too thin. Selling a component developed through research can strengthen expertise. Taking on consulting assignments unrelated to the product, by contrast, can conceal a lack of progress. Funders must examine operating margins, cash burn and the funding needed to reach the next milestone separately: an order book is no substitute for available cash.
Trust is also built through bad news
Useful reporting should show what is progressing, what is stalled and what would prompt a change in strategy. Independent validation, even partial, is worth more than another superlative. Commercial confidentiality does not prevent a company from publishing an evaluation method. For investors, this discipline helps distinguish a surmountable difficulty from an assumption that has become shaky.
What now? Looking ahead to September 2026, the healthiest scenario would be a sector less obsessed with setting a date for a revolution and more focused on successive demonstrations of progress. Specialized markets could fund part of the journey, without guaranteeing the arrival of broadly profitable quantum computing. The companies that endure may be those able to say all three things at once: here is our ambition, here is what we have demonstrated, and here is what we still do not know.
