A Vendor's Addressable Market Is Not the Market

Two sets of numbers, one year, and they cannot describe the same market

In 2025 QIAGEN put the total addressable market for quantitative PCR at approximately USD 2.7 billion. In the same year Roche reported CHF 2,527 million of Molecular Lab revenue, which converts to roughly USD 3,048 million at the 2025 annual average rate of 1.2062. Both numbers are honestly reported, both are drawn from disclosures a reader can check, and one of them is larger than the other’s entire universe. At least one of them is answering a question nobody asked.

The resolution is not a restatement or an arithmetic error. It is a definition, and the definition turns out to be the most commercially useful thing in the category. QIAGEN’s addressable market is open qPCR chemistry: reagents and consumables a laboratory buys in order to run its own assays on its own instruments. That market structurally excludes closed cartridge systems, where instrument, consumable and assay are sold as one unit and the buyer cannot substitute the chemistry at all. Closed cartridge systems are where Roche, Cepheid and bioMérieux earn the bulk of their molecular revenue.

So a figure describing open chemistry excludes the largest revenue pools in the market it appears to be sizing. That is not a vendor being misleading. It is a vendor correctly describing the portion of the world its catalogue addresses, which is exactly the right thing for a vendor to do and exactly the wrong thing for anyone else to adopt.

This is the defect the PCR and digital PCR market report was built to avoid. Its base of USD 10.42 billion in 2025 rising to USD 15.40 billion in 2033 at 5.00 percent is constructed from ten attributable revenue pools plus a stated residual, not adopted from any supplier’s estimate of what it could sell.

What a total addressable market is actually for

A TAM has a job, and the job is not measurement. It is prioritisation. It exists to tell a sales organisation which budget lines are contestable with the product it has, which is why every TAM is bounded by the vendor’s own portfolio rather than by the category’s economic activity. A reagent company’s addressable market excludes every dollar spent on an instrument it cannot run on. A closed-system company’s addressable market excludes every dollar spent on open chemistry. Neither is wrong, and neither is a market size.

Adopting one has two consequences, and they compound. The level is understated by the entire installed base of that supplier’s competitors. The growth rate is borrowed from a different business. Open qPCR chemistry tracks research funding, laboratory counts and the willingness of a laboratory to validate its own assay; its price declines because the reagent is substitutable. Closed cartridge PCR tracks cleared menus, placed instruments and payer coverage; its price holds because the menu is the product. Applying the first growth rate to a market that is mostly the second, or the reverse, produces a forecast that cannot be delivered by any participant.

The pattern is not specific to PCR. The diagnostic tests market report runs into the same problem at a larger scale, where a category defined by technology, by setting and by purchaser produces three different market sizes from the same underlying activity.

The chemistry is identical and the business is not

The sharpest evidence that this is a mix question rather than a technology question is that the same polymerase chain reaction produced wildly different outcomes at three companies in the same year. Roche’s Molecular Lab grew 4 percent at constant exchange rates. Abbott’s molecular line ran at minus 1.2 percent organic. QIAGEN’s PCR product group fell 9 percent at constant exchange rates in the fourth quarter.

Scale does not explain it either, except as a symptom. Roche’s molecular franchise is about 5.9 times Abbott’s and grew five points faster in the same twelve months. The difference is what sits inside the line: a menu of cleared assays on installed cartridge instruments, defended by reimbursement, versus a chemistry business selling into laboratories that can switch supplier on a purchase order.

China makes the point in a form that cannot be argued with. Roche’s Asia-Pacific revenue fell 12 percent at constant exchange rates on repricing, and the drag landed on the molecular line rather than on the pathology line, which grew 14 percent. Repricing is a mechanism that reaches tests with a code and a rate. It does not reach a research reagent sold to a laboratory that sets its own price. What a company earns in a repricing cycle is a function of which of the two businesses it is in.

Five per cent is arithmetic, not a view

The category rate of 5.00 percent is not a judgement about enthusiasm. It is the weighted outcome of two very different segments. Digital PCR is 6.33 percent of the base and grows at roughly 15 percent, which contributes 0.95 percentage points. Quantitative PCR is 93.67 percent of the base and grows at roughly 4.3 percent, which contributes 4.06 points. The total is 5.01 points, rounded to 5.00.

That decomposition is worth internalising because it forecloses a common error. Any forecast of high single digits for PCR is a forecast that digital PCR has become the market rather than a rounding item inside it. For that to happen at 15 percent growth, digital PCR would need the better part of two decades. Anyone publishing a mid-teens rate for the category as a whole is either sizing a digital PCR market and labelling it PCR, or has taken a supplier’s aspirational number and applied it to a base it was never meant to cover.

The base year is a trough, and that is the only reason the rate holds

A 5.00 percent forecast is only defensible if 2025 was an unusually weak starting point, and it was, for three reasons that all sit outside the technology.

China repriced molecular testing, which cost Roche 12 percent of its Asia-Pacific revenue at constant exchange rates. Academic capital spending contracted: Bio-Rad’s life science segment declined 1.3 percent at constant exchange rates, and QIAGEN stated plainly that digital PCR instrument placements came in below its own expectations for exactly that reason. And the COVID testing tail completed: Abbott’s COVID testing revenue fell from USD 747 million to USD 297 million.

None of those three is a statement about whether PCR is a good technology. All three are policy or funding events with a finite life. The forecast requires that they stop dragging. It does not require any of them to reverse, and the report says so rather than assuming a recovery it cannot evidence. That is what separates a 5.00 percent forecast from a 9 percent one: not optimism about the science, but an assumption about how long a funding contraction lasts.

The concentration is one judgement, and it is in the table

Roughly a quarter of the base is a single attribution decision, and hiding it would make the whole number untrustworthy. Roche’s Molecular Lab reported CHF 2,527 million at 4 percent constant-currency growth. Converted, that is USD 3,048 million. No participant discloses PCR as a revenue line, so the question is what share of Molecular Lab is PCR, and the report applies 80 percent. That yields USD 2,438 million, or 23.4 percent of the base.

A 70 percent split gives USD 2,134 million. A 90 percent split gives USD 2,743 million. The full range moves the category total by about USD 300 million either way, which is under 3 percent of the base, and it is stated in the body rather than buried in an appendix. The residual for unlisted and regional suppliers is USD 705 million, 6.8 percent of the base, which is the other check that matters: if the named pools genuinely cover the market, the unexplained remainder should be small. Here it is.

What to do with a number like this

Read the construction before the level. Ask which pools are inside it, what was excluded and why, and how large the residual is. A market size with no stated residual is a market size with an unmeasured hole in it, and the hole is usually where the growth is assumed to come from.

Then ask which growth rate the number is carrying, because that is where a supplier’s figure does the most damage. The level being wrong by a third is a problem a reader can see. The rate being borrowed from a different business is a problem that only appears three years later, when the forecast has been delivered by nobody. The companion piece on reagent economics follows that second problem into the instrument channel, where the same mix question determines whether a placement is revenue or a cost.