Nearly every medtech pitch deck includes a regulatory slide. It usually shows a timeline of two or three years, a box for FDA submission, and a dollar amount next to it. That number is often just the user fee, and it’s usually the smallest cost in the whole plan.
The difference between what’s shown on that slide and what actually gets spent is where much of the money goes. Founders with experience plan for costs that never show up on a fee schedule, like a second study requested by the agency, hiring for a permanent quality system, or the months between clearance and the first sale. First-time founders often discover these expenses one by one, usually later than they would have wanted.
Here’s what those obvious costs don’t cover.
The Fee Schedule Is Actually The Least Expensive Part Of The Process
FDA publishes exactly what it charges. For fiscal year 2027, FDA's medical device user fee rates put a standard 510(k) at $28,653, a De Novo classification request at $191,020, and a premarket approval application at $636,732. Qualifying small businesses pay roughly a quarter of those amounts. Every registered establishment also owes $13,785 a year just to stay listed, whether or not it shipped a single unit.
These numbers are more important as signals than as actual expenses. The real difference between a 510(k) and a PMA isn’t just $608,000 in fees. It’s the difference between simple bench testing and a multi-year clinical program with sites, monitors, and a statistician. If a founder plans for the cheaper route but ends up on the expensive one, they don’t just lose a fee—they lose a year and a round of funding. The FDA offers a free Pre-Submission program that gives written feedback on the pathway and study plans before enrollment starts. Skipping this to save eight weeks can end up being the most costly decision in the industry.
Your Predicate Is A Liability Decision
For any device going through the 510(k) process, the predicate device is more than just a regulatory shortcut. It also means inheriting a risk profile, and that risk can be measured.
A JAMA analysis of every 510(k) device subject to a Class I recall between 2017 and 2021 found that 44.1 percent had been cleared using a predicate that had itself been recalled at Class I. Compared with matched controls cleared on recall-free predicates, devices cleared on recalled predicate devices carried 6.40 times the risk of a Class I recall of their own. The median recall event in that sample covered 9,345 units, and the median device reached its first Class I recall 7.3 years after clearance.
Seven years is longer than most seed-stage timelines, which is why the cost often falls to whoever owns the company later. Checking a predicate’s recall history only takes an afternoon using public FDA databases. Finding out through a field action, though, means needing a recall budget, a regulatory consultant, and dealing with possible lawsuits.
Clinical Validation Is A Second Program, Not A Second Phase
Bench testing shows that a device works as designed. But it doesn’t show whether the results matter for patients, and that’s often where submissions get stuck.
The important decisions are made before collecting the first sample. For IVD clinical validation, you need to consider the intended use, the right comparator or reference standard, whether the study population matches the real-world users, and if endpoints and statistical thresholds are set in the protocol ahead of time. If you use a convenient but unrepresentative population, you might get good data that doesn’t support a useful claim. If your study is too small, the results may be too uncertain to support any claim, and by the time you realize it, enrollment is over, and the money is gone.
The same discipline applies as with any early-stage product. It’s better to prove your product’s value before scaling up than to build a company on assumptions. In medical devices, learning too late means repeating a study, not just changing direction, and repeat studies can take years.
The Data Layer Nobody Puts In The Model
Clinical data infrastructure gets treated as an administrative expense until it isn't. Paper diaries are filled out in the parking lot before a site visit, and responses are entered by hand into a database. If two systems can’t communicate, someone has to reconcile the data and a monitor has to check it.
Switching to electronic patient-reported outcome data changes how things work. Each entry is time-stamped as soon as it’s submitted, and responses go straight into the study database without needing transcription. You can also see whether participants are following the protocol while the study is ongoing, rather than waiting for paper diaries at the next visit. This last point is where the real costs come in. If you spot a compliance issue in month three, it’s just a phone call. If you find it at database lock, it could mean a protocol amendment or a debate with a reviewer about data quality.
Amendments are expensive. A study from the Tufts Center for the Study of Drug Development, published in Therapeutic Innovation & Regulatory Science, found that 57 percent of protocols needed at least one major global amendment, and 45 percent of those were considered avoidable by the sponsors. The median direct cost was $141,000 for a Phase II protocol and $535,000 for a Phase III. While this research is about drug development, the same issues apply to any regulated study. The cost of the amendment is only part of the problem; the delay it causes is the other.
The Quality System Becomes Permanent Overhead
On February 2, 2026, FDA's Quality Management System Regulation replaced the old Quality System Regulation at 21 CFR Part 820, incorporating ISO 13485:2016 by reference. The same day, the agency retired the Quality System Inspection Technique it had used for decades and moved to a new inspection program.
For companies that set up their quality system under the old rules, the transition means updating procedures, rewriting documents, and retraining staff. None of this creates products to sell. The new standard is also a licensed document that must be bought before anyone can read the requirements. The work doesn’t end after the transition, either. Internal audits, corrective and preventive actions, supplier controls, complaint handling, and design change control all become ongoing responsibilities with permanent costs, not just projects that end after submission.
Clearance Is Not Adoption
The most commonly underestimated cost comes after regulatory approval. The device is cleared, but no one is buying it yet.
Selling to clinical settings means facing another round of evaluation with different standards. Practices considering new equipment look at factors like how often it will be used, inventory costs, staff training, and whether patients will return. The financial factors for adopting new treatments are often unrelated to the evidence that satisfied the FDA. Payers do their own review, and the evidence that proves substantial equivalence for regulators often doesn’t show medical necessity for insurance coverage.
Securing a CPT code, making the health-economic case, and navigating payer policies is a separate, multi-year effort that needs its own team. Companies that only budget for regulatory clearance, not for coverage, can end up with a cleared device, no sales pipeline, and just six months of cash left.
Regulatory fees are easy to incorporate into a financial model because they are available as published dollar amounts. The harder costs to anticipate are those created by decisions made months or years earlier: choosing the wrong pathway or predicate, designing a study that must be repeated, underestimating permanent quality costs, or waiting until clearance to consider reimbursement.
Those risks cannot always be eliminated, but they can be identified before they become expensive. A few questions can reveal where a regulatory budget still relies on assumptions rather than a real plan.
The Founder’s Checklist
Six questions worth answering in writing before the regulatory line item goes into a model:
-
Has the pathway been confirmed with the FDA through a Pre-Submission, or is it assumed internally?
-
What is the recall history of every predicate in the lineage being cited?
-
Are endpoints, thresholds, and power calculations locked in the protocol before enrollment?
-
Does the study population match the population the device will actually be used on?
-
What does the quality system cost annually once the submission is behind you?
-
Who owns reimbursement strategy, and when does that work start relative to clearance?
The point is not to predict every expense perfectly. It is to identify the assumptions that could become six-figure problems while they are still inexpensive to question. Getting those answers early may add time and expertise to the front end of the process, but it is considerably cheaper than discovering them after the study is complete, the submission is filed, or the runway is already running short.
