The two recovery curves.

Two hundred extra days of FDA review is the cheapest commercial runway anyone in this industry will ever be handed, and almost every company receiving it this year will spend it waiting by the letterbox.

BTIG pulled the FDA's device numbers for the first half of this year and the shape of them is worth sitting with. More devices came through than in the same period of 2025, 23 original premarket approvals against 13, with 1,669 510(k) clearances and the total trending up around 2.5% on the year. By the measure most companies actually watch, the agency is having a strong year.

The other number moved in the opposite direction. Average time to original premarket approval went from roughly 402 days last year to nearly 599 days in the first half of this one, close to 200 additional days on the clock while the output figure climbed. The 510(k) average moved far less, up seven or eight days to around 156, and de novo timing rose by a little over 8%.

BTIG were explicit that a handful of outliers dragged the premarket average upwards, which means the honest description of what has happened is not that everything got slower. Most submissions still behave roughly as they used to, a small number now overrun by a margin that would have been difficult to imagine two years ago, and there is no dependable way to know in advance which one you are holding. The middle held, and the tail got fatter.

A workforce event underneath the numbers

Underneath both numbers sits a workforce event. Between September 2024 and January 2026 the FDA lost about 21% of its people, more than 4,400 of them, and those who stayed inside the device centre have been carrying the load since, with attorneys and consultants who work closely with the centre reporting the loss of experienced leaders and communication that has become harder to read. The agency is now bringing 2,200 back, with around 600 onboarding.

It takes two to three years to train a reviewer.

That is the mechanism, and it is worth being precise about how it runs. A reviewer two years into the job handles a routine submission with a clean predicate about as competently as one with fifteen years behind them, and the distance between those two people shows up on the file that does not fit the pattern, the novel technology, the ambiguous predicate, the question that has to be escalated because nobody in the room has seen it before. Attrition does not slow an agency down evenly, it slows down the unusual.

The last time a workforce was cut all at once

If that sounds like a condition that clears once the rehires land, it is worth knowing how the last one went. In August 1981 Ronald Reagan fired 11,359 air traffic controllers, and the thing everybody remembers about it is that the planes kept flying. Supervisors and military controllers filled the towers, the system ran at reduced capacity for a few months, airlines trimmed their schedules, and before very long the departure boards looked much as they always had. What almost nobody outside aviation registered was that the FAA was still hiring retired controllers back into areas of critical shortage more than a decade afterwards, and the airlines expanding hardest through the deregulation years found the rebuilding of that workforce to be an obstacle they had never written into a plan, because their route economics quietly assumed a system running at a capacity it would not reach again for years.

Nobody had misled them. The input had changed underneath the plan while the visible output carried on looking entirely normal, which is the most expensive way an assumption can fail.

So the 2,200 do not close this. Headcount can be restored inside a year and judgement cannot, which makes the fat tail a condition to build around rather than a delay to wait out.

Who the fat tail lands on

It is not randomly distributed either. It sits over the submissions doing something genuinely new, and the companies filing those are, overwhelmingly, the ones with the least balance sheet to absorb the wait. J&J took the Ottava authorisation this year and Edwards took approval for its surgical tricuspid valve, and neither of those organisations changes the character of a board meeting over an extra 200 days, whereas a company holding 14 months of runway changes everything about the next one.

That is where the commentary stops, and it is not wrong. It is just not the useful part.

The useful part is that everybody in that queue has been handed the same thing, which is time, and the only question that matters is whether they hired somebody capable of spending it.

The function hired against a date that doesn't exist

Because the function carrying this is the one always hired against a date that does not yet exist. Pre-commercial teams are recruited on a forecast clearance rather than an actual one, because a commercial leader needs a quarter or two to build the territory model, hire beneath themselves and get the first accounts warm, and starting them on the day the letter arrives is already late. So the question every board in this position asks is a timing question. Hire now and risk paying someone to wait, or hire later and arrive at first revenue with no engine. Both answers are wrong, and they are wrong for the same reason, which is that timing was never the variable.

A commercial leader recruited to launch a product is dead weight until there is a product. A commercial leader recruited to build the conditions the launch depends on is fully occupied from the first week. The payer pathway does not need a clearance letter to be designed, and neither does the evidence the payer is going to demand, or the map of the fifteen accounts that decide whether year one looks national or regional, or the read on which competitor is stuck in the same queue behind you and does not yet know it.

One of those two people spends 200 days becoming an expensive question at a board meeting, their internal credibility spent on a launch that has not happened. The other spends them arriving at clearance with the reimbursement argument already made and the accounts already warm, against a competitor who spent the same period waiting.

That is a different specification, and it is a different person. It is not the launch VP with three product introductions behind them. It is the one who has built a commercial position before there was anything to sell, which is a much rarer thing and almost never what the job description asks for.

The fat tail does not punish the companies that hired early. It punishes the companies that hired narrowly.

Your regulatory date is a mean and you cannot fix that. Your job specification is entirely within your control and almost nobody is looking at it.

Which one is your launch actually exposed to?