2026 tariffs and RT manufacturer margins: where the recoverable margin actually is

What the 2026 tariffs actually cost, with sources

Start with figures that can be checked, because most of the numbers circulating in this conversation cannot be. The UNC Center for the Business of Health reports that duties on medical devices range from 10 to 50 percent depending on country of origin, and that 62 percent of the medical devices used in the United States are imported. That is the shape of the exposure: not one rate, and not a niche one.

At company scale, MedTech Dive reported in April 2026 that individual large medtech manufacturers were carrying annual tariff-related costs in the range of $200 million to $500 million. An accessory manufacturer’s exposure is far smaller in absolute terms and far more concentrated in relative terms, because it sits on a narrow set of imported inputs behind the high-volume consumable lines that carry the revenue.

That figure describes a peak that has since moved. Executive Order 14389, signed on 20 February 2026, ended the additional duties that had been imposed under the International Emergency Economic Powers Act. It expressly left Section 232 and Section 301 untouched, and on 24 July 2026 a new Section 301 action took effect adding 12.5 percent on products of China as part of a 60-economy forced-labor programme. Separately, Commerce’s Section 232 investigation into imports of medical equipment and devices, opened on 2 September 2025, is still listed as open with no published determination.

The useful lesson there is not a rate. It is that the rate moves, is litigated, and is sometimes refunded a year after it was paid. A manufacturer who builds its entire response around one number has to rebuild it every time a court or an executive order changes that number. The parts of the response worth building are the ones that hold whatever the rate turns out to be.

Tariff schedules move. Rates depend on HTS classification and country of origin, and the figures above are what the named sources published as of this briefing’s date. Confirm your own exposure with customs counsel against your own bill of materials rather than against an industry average.

The 2026 tariff exposure for an RT accessory manufacturer, precisely framed: duties on medical devices run from 10 to 50 percent depending on country of origin, against a US market where 62 percent of devices in use are imported (UNC Center for the Business of Health). Individual large medtech manufacturers faced annual tariff-related costs of $200 million to $500 million, as reported in April 2026 (MedTech Dive). For an accessory maker the absolute number is smaller but the concentration is higher, because the exposure sits on a short list of imported inputs behind the highest-volume consumable lines. And the rate itself is unstable: the emergency-powers duties were ended by executive order in February 2026 and much of what they collected became refundable, while a new Section 301 action added 12.5 percent on products of China from 24 July 2026 and a Section 232 medical-device investigation opened in September 2025 remains undetermined. The open question is therefore not what the rate is. It is which part of the business can give margin back regardless of the rate.

The sector absorbed it, and absorption has a limit

The most useful finding in the MedTech Dive reporting is not a number. It is the posture. Manufacturers, it found, are largely not passing prices on to customers and are instead hunting efficiency across their operations. Most of the industry ate it.

That was a reasonable first move, and the reasons are worth naming honestly. Health systems are carrying their own cost shock: tariffs are expected to increase hospital expenses by at least 15 percent in the near term, according to the American Hospital Association’s resource and materials management association. Contract vehicles constrain when a price can move. And a price increase presented without evidence does not land as a price increase, it lands as an invitation to review the whole line. A manufacturer who raises a price and cannot say what the buyer is getting for it has opened a competitive evaluation they did not intend to open.

But absorption is a decision to fund the tariff out of margin, and margin is finite. The second move is the one that matters, and for most accessory manufacturers it is not a bolder price increase. It is recovering the margin that is already inside their own channel and has simply never been measured: the orders that arrive wrong, the accounts nobody is really covering, and the revenue that is concentrated in a single import regime.

Why absorption is a first move and not a strategy: MedTech Dive’s April 2026 reporting found manufacturers declining to pass tariff costs to customers and pursuing operational efficiency instead. Absorption buys time and protects the customer relationship, but it funds the tariff directly out of margin, and it compounds for as long as the regime lasts. The manufacturers who eventually reprice successfully will be the ones who can show a buyer what they are paying for, line by line, rather than announcing a percentage. Evidence is the thing that makes the second move survivable.

Repricing is an evidence problem before it is a pricing problem

Ask a manufacturer what happened to their line last quarter in the channel and you will usually get an answer about units shipped. That is the whole difficulty. Once product leaves for a channel, most manufacturers stop seeing it: not which institutions bought, not how quickly anyone responded when something went wrong, not how many units came back or why. The information arrives months later, in aggregate, if at all.

This is not an argument against distribution. Distribution is how accessory manufacturers reach a fragmented buyer base, and no software replaces a channel that carries inventory and answers the phone. It is an argument against opacity, which is a different thing and a fixable one. A manufacturer who cannot see the channel cannot answer the three questions a defensible price move depends on: which accounts are buying enough to notice a change, which ones are absorbing service cost that never shows up in the price, and which product families are generating returns that quietly erase the margin on the ones that ship clean.

OncoSource is a distributor, built as software. We carry a curated line of manufacturers rather than a catalog of everything, we are non-exclusive by design so your existing book stays entirely yours, and the difference is that every step is instrumented. Compatibility is resolved before an order is placed. Support tickets carry a first-response clock and a resolution clock, and a breach of either is recorded permanently. Returns are captured with a reason code. All of it is reported back to you, about your own line, in a manufacturer portal rather than a monthly summary.

One rule governs every number on that surface. Any measure with fewer than three distinct buying institutions behind it returns no number at all rather than a flattering one. That floor is enforced in the code, not in a policy document, and it cuts against us more often than it helps: several fields read empty until the orders exist to fill them, and we would rather tell you that here than have you discover it in a demo.

What an instrumented channel gives a manufacturer under cost pressure: evidence, on its own line, in time to act on. Which institutions ordered, how fast anyone responded when something went wrong, how many units came back, and whether they came back because the product did not fit the equipment in the room or because the order shipped wrong. Those two failure modes have two different fixes, and a single blended return rate hides both. The distinction is only available if the channel is built to record it, and the number is only worth anything if the channel refuses to publish it below a real sample. Ours returns nothing at all below three distinct buying institutions.

The cheapest margin in the building is the order that does not come back

For accessory lines, a meaningful share of margin leakage is not pricing at all. It is configuration. The wrong index, the wrong thickness, the wrong fit for the couch that is actually in the room. The return that follows costs the manufacturer the unit, the freight both ways, the handling, and a measure of credibility with the department. It is also usually invisible on a report, because it is not counted separately from any other return.

The fix is upstream of the return. Compatibility is checked against the buying institution’s actual installed base before the order is placed rather than discovered after it ships. Ordering is built to run inside the buyer’s own procurement system: the requisition is built there over the cXML PunchOut standard the platform implements (Oracle, SAP Ariba, Workday, Coupa, Infor Lawson, JAGGAER and Dynamics 365) and lands with us electronically, so nothing is retyped out of a PDF. Fewer configuration errors upstream is the only durable way to get fewer returns downstream.

Said plainly about where we are: those ordering rails are built and the PunchOut integration is implemented, and they are switched on when the first manufacturers are ready to transact. We are deliberately early and working with a small first group of manufacturers rather than a running book of business, and we would rather you hear that from us than find it out later.

Why error reduction is the fastest margin available under tariff pressure: a price increase requires a negotiation, a contract window, and a buyer willing to absorb it. A return that never happens requires none of those. For accessory lines, configuration error is a recurring and largely unmeasured cost: the unit, the freight both ways, the handling, and the credibility. Resolving compatibility against the institution’s actual installed base before the order is placed, and coding every return by reason so “did not fit” is counted separately from “shipped wrong”, turns an invisible cost into a tracked one. You cannot reduce what nobody is counting.

The accounts a field team cannot reach

The second place recoverable margin hides is coverage. Mid-market hospitals, independent cancer centers, and freestanding outpatient radiation therapy facilities buy in smaller, more frequent orders than a large health system does, and the per-account revenue rarely justifies a rep’s windshield time. They are not accounts where a manufacturer is losing. They are accounts where a manufacturer was never present, which is a harder thing to see on a report.

These buyers also research differently. A freestanding center has no value-analysis committee with a nine-month calendar and usually no dedicated accessories buyer. Procurement runs through a small administrative team, decisions close in days rather than quarters, and the default motion is digital: search, compare, request a quote. A manufacturer who is not represented at that comparison is not losing the evaluation, they are absent from it.

That is the segment the channel is built for, and it is additive rather than competitive: these are, by construction, the accounts your field team is not working. We read observed public procurement activity at the level of geography, segment and institution, and we use it with you to build the target map for your line: which institutions sit in your competitive set, which are active in the public record, and which have never seen your products. That is a view of publicly observable buying rather than of the whole market, and we would rather say so than imply we see more than we do. We do not publish anyone’s contract pricing, and we do not promise you a price benchmark we cannot stand behind.

A second market is the one hedge a tariff cannot follow

Every lever above operates inside one country. A tariff is a tax on importing into a specific market, and it does not follow your product into a market you are not selling in yet. For a manufacturer whose revenue sits almost entirely behind a single import regime, geographic concentration is itself the exposure, and diversifying it is a hedge rather than an ambition.

The reason most accessory makers never do it is not commercial. It is that a market where you hold no clearance and have no legal presence is a regulatory and infrastructure problem before it is a sales problem. That is the second thing OncoSource does, and we run it before selling a single unit into the market. We manage the clearance route and run the filing, and we take the roles a foreign manufacturer cannot hold for itself: US Agent in the United States, and the local registration holder or legal representative in markets that require one. In the first conversation we tell you which markets we already operate in and which we would be standing up for you. Those are different timelines and different costs, and you should know which one you are being quoted.

Two points of scope matter more than anything else in that conversation. Registration is scoped per product family, not per SKU, which is why a phased entry is affordable: most catalogs collapse into far fewer families than SKUs, and we do that collapse with you before anyone quotes anything. And in most of the Latin American markets we target, regulators operate reliance or abridged routes that lean on an existing US clearance plus ISO 13485 and a certificate of free sale, so the expensive laboratory work happens once. Brazil is the notable exception and runs its own dossier.

The heavy line items in a clearance are laboratory testing and agency fees. Those are pass-through and we show them to you as pass-through; we charge a scoped fee for the work, priced per family and per market. And we raise the hardest term first: before anything is filed, the agreement states what happens to your registrations if we part ways. Registration ownership on termination is the most consequential clause in any market-access arrangement, and a partner who leaves it vague is telling you something.

Why geography belongs in a tariff conversation: a tariff is a tax on importing into one market. Revenue earned in a different market is not exposed to it. For an accessory manufacturer concentrated behind a single import regime, a second market is a structural hedge, and the barrier is regulatory rather than commercial: no clearance, and no legal entity to hold one. Scoping registration per product family rather than per SKU is what makes a phased entry affordable, and reliance routes in much of Latin America mean the laboratory work is done once rather than per country. It is a slower lever than fixing returns, and it is the only one that changes the shape of the exposure instead of the size of it.

What to measure before you change anything

None of the above is a recommendation until it is measured against your own line. The inputs that decide it are these, and they are specific to the manufacturer:

• Cost of goods by product line, and how much of it moves with imported inputs. Not a blended COGS figure. The question is which lines are exposed and by how much, because the answer usually concentrates in two or three families rather than spreading evenly.
• Your return rate, split by reason. How many units come back, and how many came back because the product did not fit the equipment in the room rather than because the order shipped wrong. If your current channel cannot give you that split, that is itself the finding.
• Real coverage versus nominal coverage. Which accounts does a rep actually visit, and which appear on a territory map without anyone working them? The second group is where a non-exclusive channel is additive rather than cannibalising.
• Sell-unit correctness. Whether your catalog is represented in the sell unit a buyer actually orders in. This sounds like a formality and is not: it is the difference between a price that is right and one that is off by the size of a case.
• Revenue concentration by market. What share of revenue is exposed to a single import regime, and what a phased entry into one additional market would cost, scoped to the two or three product families you would lead with rather than the whole book.

We work through those five with a manufacturer in a working session before anything is quoted, and the honest outcome is sometimes that the recoverable margin in your line is somewhere we do not help.

The channel economics model as a decision tool: the question is not “should we add a channel” but “where is the recoverable margin in this line”, and there are only four candidate answers: price, configuration error, coverage, or geography. Deciding between them needs five inputs a manufacturer already owns or can get: exposed COGS by line, return rate split by reason, real versus nominal account coverage, sell-unit correctness, and revenue concentration by market. A manufacturer who can produce those five can make the call defensibly to a board. A manufacturer who cannot produce the second one has already found something worth fixing.

Model your channel economics

The first step is a thirty-minute call, operator to operator, to work out whether there is anything here for your line. Bring your cost structure, your return numbers if you have them, and an honest account map.

If there is something here, the next step is a working session at no cost in which we represent your catalog correctly: products, real sell units, and equipment compatibility. That one takes considerably longer than thirty minutes, and getting the sell unit right is not a formality: it is the difference between a price that is correct and one that is off by the size of a case. From there your line is staged and verified in production while still switched off, you review it before a single buyer sees it, and activation is your decision on your timing. If you are also weighing a market you hold no clearance in, that session adds one step: we collapse your catalog into product families and scope the filing and the annual cost for the two or three you would actually lead with.

Frequently asked questions

The questions RT accessory manufacturer CFOs and VPs of Sales most often ask about margin decisions under tariff pressure.

OncoSource is a distributor for radiation oncology, built as software: a curated line of manufacturers carried through an instrumented channel, plus full-stack market entry where a manufacturer holds no clearance. Tariff and cost figures on this page are attributed inline to the named public sources on the dates those sources published; rates change and depend on classification and country of origin. Nothing here is a projection of outcomes for any manufacturer.

Published September 28, 2026.