Marin Ivezic
Between 2021 and 2022, three quantum computing companies became publicly traded in the United States. IonQ, Rigetti and D-Wave each merged with a special-purpose acquisition company, a listed shell that takes a private business public without a conventional IPO, and each one now files audited annual reports and quarterly reports with the Securities and Exchange Commission. Those filings are the most useful free documents in quantum commercial diligence, and almost nobody reads them before a first vendor meeting.
They’re useful because they answer a question no pitch deck is built to answer. What does revenue actually look like in this industry, at this stage, for a company that has already survived long enough to list? Read three of them and the calibration problem largely dissolves. A private company’s numbers stop being abstract and start being comparable to something.
This article covers the commercial half of quantum vendor diligence, meaning revenue quality, milestone structure, capital plans, concentration risk, and the regulatory facts that set the size of the market. The physics and the engineering are a separate assessment. In our experience the commercial half is where quantum evaluations run thinnest, and it’s where most of the disappointments actually start. Technology risk in this sector gets discussed constantly and is reasonably well understood by anyone paying attention. Financing risk is not.
The Four Kinds of Money
A quantum company’s revenue line usually blends four sources that behave nothing alike, and a single blended figure tells you almost nothing.
Grants and public programmes. Non-dilutive funding, meaning the company receives it without giving up equity. It’s normally awarded on technical merit after peer review, so it says something real about scientific credibility. It says nothing about demand.
Research contracts. A national laboratory or a large corporate research group pays for a joint project with a defined scope. Real money and a real counterparty, but the deliverable is knowledge rather than a product, and the contract rarely renews on its own.
Professional services. Advisory work, training, algorithm development for a client’s specific problem. This is honest revenue and it keeps the lights on. It’s also headcount-bound, lower margin than the business the company is pitching, and it competes for the same engineers who are supposed to be building the product.
Product and access revenue. Cloud access hours, on-premises systems, software licences, support contracts. This is the only line that demonstrates a market exists.
None of the first three counts against a company. An early-stage quantum business funded largely by grants and research contracts is behaving sensibly. The problem is presentation, when four kinds of money get reported as one number and then described in the vocabulary of product-market fit.
A worked example of how that resolves. A deck claims eight customers, two of them Fortune 500. Diligence turns that into two paid cloud-access subscriptions with annual renewal, three seats in a government-funded consortium where the government pays and the members participate, two unpaid proofs of concept, and one memorandum of understanding, a written statement of intent to work together that carries no obligation to buy anything. Eight is accurate. Two renew.
Three requests will settle most of this. Ask for last quarter’s invoiced revenue split across the four categories above. Ask which contracts carry a renewal date inside the next twelve months, and what the historical renewal rate has been. Then ask which named partner has issued a purchase order rather than a letter of intent, which is a non-binding statement that a buyer expects to buy at some point. That gap between the two documents is what most quantum commercial reporting blurs.
The Roadmap Is a Financing Document
Public roadmaps in this industry serve two audiences at once, customers and capital markets, and the second audience usually wins. That isn’t dishonesty. A hardware company that has to raise again every eighteen to twenty-four months needs a slide where the curve goes up, and everyone in the room knows it.
So assess milestone quality rather than milestone ambition. A usable milestone is dated, measured by a stated method, and capable of failing in public.
Weak: 2027, scale to fault-tolerant operation.
Usable: Q3 2027, two logical qubits, meaning error-corrected qubits each assembled from many physical ones, held below the error rate of their underlying physical qubits across 1,000 rounds of error detection, with the data posted as a preprint.
The second version can be checked by an outsider on a specific date. Either the preprint exists or it doesn’t. The first version can be declared achieved on almost any evidence, and often is.
Then do the archaeology. Pull the roadmap the company published two years ago, and the press releases and conference talks from the same period, and compare them line by line against the current plan. Slippage is normal in this field and should not, by itself, worry anyone. Silent slippage is the finding. Ask which dates moved, by how much, what caused each move, and whether the company told its customers and investors at the time or quietly redrew the chart.
Runway, Burn, and the Next Provable Result
Burn rate is net cash leaving the business each month. Runway is how many months of cash remain at that burn. In quantum hardware both numbers run high, because the cost base includes cryogenic systems, vacuum and laser hardware, cleanroom or foundry access, custom control electronics, and a payroll weighted toward doctorate-level staff who are being recruited by everyone else in the sector at the same time.
The question worth asking is narrow. Does the runway extend past the next milestone that an independent party could verify, with margin for that milestone to slip by a quarter or two? If the company has to raise before it can prove anything new, the next round gets priced on narrative rather than result. In a soft funding market, narrative pricing is how sound technology ends up sold cheaply or shut down entirely.
Ask what happens if the round doesn’t close on schedule. A credible management team answers with specifics, naming which programmes pause, which roles go, and how many additional months that buys. A team that answers with confidence about investor interest has not done the exercise.
Buyers have a version of this too. Payment gated against acceptance tests, with tranches released on demonstrated performance rather than on delivery date, is standard practice in sectors that fund long science and it remains underused in quantum procurement. It protects the buyer, and it also gives an honest vendor a way to prove the thing they keep being asked to prove.
Standards and Export Control as Commercial Facts
Two regulatory facts set the addressable market for whole categories of quantum company, and both are commercial before they are technical.
In August 2024, NIST published three post-quantum cryptography standards. FIPS 203 specifies ML-KEM (formerly Kyber), FIPS 204 specifies ML-DSA (formerly Dilithium), and FIPS 205 specifies SLH-DSA (formerly SPHINCS+). A fourth signature standard, FN-DSA, has not been finalised. For any vendor selling cryptographic products, the commercial question narrowed sharply the day those documents appeared. Does the product implement the standardised algorithms, and is it on a credible path to FIPS validation? A vendor still selling a proprietary scheme, or a pre-standard variant of a selected one, is selling into a market that shrinks every quarter regardless of the merits of the design. Our companion guidance on migration methodology sits at pqcframework.org.
Export control is the second. Quantum computing hardware, and some of the software that runs it, is subject to export control in several jurisdictions, and classification generally turns on device parameters rather than product names. Two consequences follow for a vendor. Some export markets require licences, which lengthens sales cycles and complicates cloud delivery, and some hiring and international collaboration becomes constrained. A management team that cannot describe how its own product is classified, and which of its target markets that classification closes, has a gap that will surface during the first government or defence procurement, which is frequently the first large procurement available to them.
A smaller point in the same family. Vendors who report performance only in proprietary metrics of their own construction are harder to compare, and harder to compare means slower to buy. Willingness to publish against common benchmarks is partly a technical signal and partly a statement about how the company expects to compete.
Concentration Risk on Three Axes
Customers. Publicly traded quantum companies routinely disclose customer concentration in their risk factors, and the disclosures are instructive reading. Ask a private company for the same figure: what share of last year’s revenue came from the largest customer, and from the largest three. Then ask what happens to the plan if the largest one doesn’t renew.
Suppliers. Dilution refrigerators, specialised laser systems, cryogenic cabling, certain isotopically purified materials and advanced packaging all come from short supplier lists, and lead times are measured in many months rather than weeks. Ask which single component has the longest lead time, how many vendors can supply it, and whether the company has ever qualified a second source. An answer that names the component and the alternative is a good sign about how the operations side thinks.
People. In a company of forty, one departure can end a programme. Ask who wrote the core control software, whether it’s documented well enough for someone else to maintain, and what the retention arrangements look like for the four or five people whose leaving would change the roadmap.
A Scorecard You Can Defend
Score the commercial assessment red, amber or green across five categories, and keep them commercial. Revenue quality. Milestone credibility. Capital adequacy. Regulatory position. Concentration.
The value of the exercise is in separating a red that’s expected from a red that’s a decision. Weak product revenue at a Series A hardware company is expected, and the correct response is to record it and move on. Capital adequacy in red is a decision, because it means the company will be renegotiating with someone before it can demonstrate anything. Regulatory position in red for a cryptography vendor is close to fatal, because the standards already exist and the company has chosen not to align with them.
Write down, next to each amber, the specific event that would turn it green and the date by which it should happen. That converts a snapshot into a monitoring plan, and it gives both sides something concrete to review at the next board meeting or the next contract renewal.
A Five-Day Commercial Sprint
When there isn’t time for a full process, five focused days will surface most of what a longer process would find.
Day one, comparables. Read the revenue disaggregation and the risk factors in the most recent annual reports of the listed quantum companies. Two or three hours, and it recalibrates everything you read afterwards.
Day two, revenue reconstruction. Request the four-way split of last quarter’s invoiced revenue and the renewal schedule. Reconcile it against the customer list in the deck.
Day three, reference calls. Two customers, selected by you from the full list rather than offered by the vendor. Ask what was paid, what was delivered, what was expected and not delivered, and whether they intend to renew.
Day four, roadmap archaeology. Old roadmaps, old press releases, recorded conference talks. Build the slippage table and take it into the next management meeting.
Day five, one expert call and a written pre-mortem. Assume the company has failed twenty-four months from now, list the plausible causes, rank them, and check which ones the material you’ve gathered can already answer.
What Quantum Changes, and What It Doesn’t
Three things genuinely differ from a conventional technology assessment. The interval between capital going in and product coming out is long enough that staying power is part of the product evaluation rather than separate from it. Performance metrics are not standardised, so comparisons have to be constructed instead of looked up. And public funding makes up a larger share of sector revenue than in most commercial markets, which means a change in national research policy is a change in demand.
Everything else holds. A purchase order is still a purchase order, a renewal rate is still a renewal rate, and a company that won’t show you its revenue composition is telling you what the composition is. Commercial diligence here isn’t a specialist discipline borrowed from physics. It’s ordinary diligence applied in a market where the ordinary signals arrive late and the technical story is loud enough to drown them out.
For the technical half of this assessment, and for the separate work of decoding vendor performance claims, the original long-form analysis on PostQuantum.com covers the physics and engineering lenses in depth.
Quantum Academy’s business-track certification programs teach this work directly, covering market assessment, vendor and technology evaluation, and building the internal case for a quantum program with defensible numbers. The current catalogue is at quantumacademy.com/.