The Bigger Line Item: What European Public Buyers Actually Commit to Keeping Medical Equipment Running — and Who Wins That Work

A Rongtao Medical industry report — July 2026

1. The line item nobody puts in the business case

Takeaway: in the European public record, keeping medical equipment running is a bigger and far more frequent commitment than buying ultrasound equipment. This is not a projection — it is the award notices.

We pulled the EU's Tenders Electronic Daily record for two things: the ultrasound-equipment CPV codes, and the codes public buyers use when they contract for repair and maintenance of medical equipment. The window is 2016 to mid-2026 and contains 29,979 notices, of which 12,831 are contract-award notices — the ones that record money actually committed rather than money contemplated ¹.

Three scopes come out of that, and keeping them separate is the whole basis of an honest comparison:

Source: EU TED contract-award notices — Rongtao Medical analysis, July 2026 ¹.

Two notes on method, because they change how the numbers should be read.

The totals are trimmed at the 99th percentile inside each scope. A small number of framework agreements in this record run to nine and even ten figures — a single notice covering every category of equipment for a whole national health service for several years. Left in, they swamp any sum and turn the analysis into a story about four outliers. Trimmed, the picture is stable and the medians do the work.

The service scope is broader than ultrasound, and we are not going to pretend otherwise. CPV 50421000 is repair and maintenance of medical equipment, full stop. It is not possible to isolate ultrasound service inside it, because buyers do not tender it that way — they tender the fleet. So every service figure in this report describes medical equipment generally. The equipment figures, by contrast, are genuinely ultrasound-specific: CPV 33112200 (ultrasonic unit) and 33112340 (echocardiograph).

With that scoping in place, the headline is simple. Four service awards for every ultrasound-equipment award, and more than twice the committed euros. For anyone whose business plan treats service as an afterthought attached to a sale, the record says the afterthought is the larger business.

"But I only work on imaging"

The obvious objection to a medical-equipment-wide figure is that an imaging service company cannot address all of it. Fair — and the classification gives us a way to answer it, if not a perfect one. Inside the service scope, the awards split across four CPV codes, one of which is imaging-specific ¹:

So the explicitly imaging-scoped slice is 944 awards and €1.21bn — 15.6% of the service awards by count, and still 63% as many awards as the entire ultrasound-equipment scope (944 against 1,496), with a median award of €483,030, slightly above the general medical-equipment median.

Two things follow. First, even the narrowest, most conservative reading — imaging-only service, ignoring every general medical-equipment contract — is a market of nearly a thousand publicly awarded contracts worth over a billion euros in eleven years. Second, and more practically: the general code is where most of the work is, and a company that can only service imaging is choosing to compete for 15.6% of the awards. The buyers who tender under 50421000 are tendering a mixed fleet, and they are the larger opportunity for anyone able to cover one.

The imaging-specific code concentrates in Romania (323 awards), Spain (227), Poland (91), France (70), Czechia (53) and Portugal (46) ¹ — broadly the same countries that dominate the general service picture in section 6.

2. Why these numbers exist at all

Takeaway: TED only contains contracts above the EU publication thresholds — which means the small-contract share measured later in this report is a floor, not a ceiling.

Before drawing conclusions from a procurement database, it is worth knowing what forces a contract into it. Under Directive 2014/24/EU a public contract must be advertised EU-wide, and therefore appear in TED, once its estimated value excluding VAT crosses a threshold. For 2026–2027 those thresholds are ⁵:

Set by Commission Delegated Regulation (EU) 2025/2152 of 22 October 2025.

That raises an obvious question about the figures in section 4, where 12.1% of service awards come in under €50,000 — well below any of those thresholds. There are two documented reasons a below-threshold award appears in TED anyway. An above-threshold procurement can produce only below-threshold awards: a tender for a hospital group's imaging maintenance, split into fourteen lots, may award every individual lot below the threshold even though the whole exercise was well above it. And contracting authorities may publish a below-threshold award notice voluntarily, for transparency ⁶.

The consequence runs in the direction that matters for this report, and it is worth being explicit about because it is the opposite of a caveat:

Most genuinely small service contracts never enter TED at all. They are procured nationally, below threshold, through each member state's own portal. So when section 4 reports that nearly a third of the visible market is under €200,000, that is the share visible inside a database designed to capture large contracts. The real small-contract market — the one a regional service company would actually live on — is larger than this record can show. We cannot size the invisible part, and we will not pretend to.

One coincidence is worth flagging to any reader who works with these thresholds daily: the 25th percentile of service awards in our record sits at €152,400, between the central and sub-central thresholds, and the €200,000 band boundary used in section 4 sits just under the €216,000 sub-central threshold. The distribution we can see is shaped by where the publication rule bites.

3. It is not a one-year artefact

Takeaway: the ratio holds in every year of the record, and it has been widening since 2021.

A single aggregate can hide a one-off. This one does not. Split the award notices by year and the service side leads in all eleven years ¹:

The absolute volume of service awards has grown from 153 in 2016 to 1,025 in 2025 — nearly seven-fold — while ultrasound-equipment awards grew from 40 to 198. Some of that is TED's own coverage improving over the decade, which is why the ratio is the honest measure rather than the raw count. The ratio is up too: taking each period as a whole rather than averaging thin years, it ran 3.4× across 2016–2020 and 4.5× across 2021–2025.

The money says the same thing as the count

A count of notices could still describe a market of many small service contracts alongside a few enormous equipment purchases. It does not. Sum the committed value inside each of this report's two scopes, year by year, and the service side is the larger number in every one of the eleven years ¹:

Values are trimmed at the 99th percentile within each scope, as everywhere in this report.

The service share moves in a band — 85% in the thin early years, a trough at 59% in 2020, back to around 70% since — but it never drops below half. That stability matters more than any single year's figure. A ratio that survives a pandemic, a procurement-rule change and eleven budget cycles is describing a structural feature of how health systems buy, not a moment.

It is worth naming the 2019–2020 trough, because it is the one period where equipment closed the gap: committed equipment value rose from €297m in 2018 to €538m in 2020 while service value plateaued around €765m. That is what an equipment-purchasing surge looks like in this record. It lasted two years, and the service line resumed its climb immediately afterwards, reaching €1.62bn in 2025 — more than double its 2019 level.

That direction is not mysterious. Equipment purchases are lumpy, grant-funded and postponable. Maintenance is not postponable, because a scanner that is down stops a clinic list today. When budgets tighten, the purchase slips a year and the service contract gets renewed.

4. The market is mostly small contracts

Takeaway: the €5m framework agreements dominate the conversation and 8% of the awards. Nearly a third of the market is under €200,000 — which is the part a regional company can actually bid for.

Aggregate market size is a vanity number if the whole thing is locked up in a handful of national frameworks. So the more useful cut is the distribution of individual award sizes ¹:

Read the top two rows together: 29.8% of service awards are under €200,000. More than half — 52.1% — are under €500,000.

The medians tell the same story from another angle. The typical medical-equipment service award is €455,264; the typical ultrasound-equipment award is €629,900 ¹. Individually, service contracts are smaller. There are simply four times as many of them.

This matters more than the headline total, because it answers the question a distributor or regional service company is actually asking: is there anything here I could win? A market whose typical unit is a €455,000 multi-year service contract, with a third of its awards under €200,000, is a market with an entry ramp. A market of €50m national frameworks is not.

The quartiles are worth sitting with for a moment, because they are what a bid actually has to be priced against. A quarter of service awards come in under €152,400 — and remember that a service award typically runs for several years. Spread across three years and a mixed fleet, that is a contract where a single unbudgeted whole-assembly replacement can consume the margin for the entire term. It is precisely the band where the difference between replacing a board and repairing a board stops being a technical preference and becomes the difference between a profitable contract and a loss-making one.

The same logic runs the other way at the top. Above €5m, contracts are usually multi-site frameworks with staffing, uptime guarantees and penalty regimes that require a service organisation, not a service capability. That band is 8.4% of awards and it is not where a company enters this market. It is where a company that has already won thirty small contracts eventually ends up.

What one of these contracts looks like

Abstractions are easy to argue with, so here are real awards from the record, all from 2026, all in the band a regional company could bid for ¹:

These are individual hospitals and regional health authorities, not national frameworks, and the winners across this band are a genuine mix: manufacturer subsidiaries appear, and so do independent regional service companies with no manufacturer name at all. There are 507 awards in this record between €40,000 and €260,000 that name their winner — a large, concrete, checkable population of exactly the contracts this report is about.

And remember what section 2 established: this distribution is the visible one. The bands below €216,000 are populated by lot-level awards and voluntary publications inside a system built to capture larger contracts. Whatever the true shape of the small-contract market is, it is at least this large.

5. Who wins it today — and the assumption that turns out to be wrong

Takeaway: the folklore says imaging service belongs to the OEM. In the named public record, somewhere between 63% and 71% of medical-equipment service awards went to someone else, across 991 distinct company names.

Of the 6,050 service awards, 1,981 (32.7%) name their winner. Within that named subset ¹:

• 585 awards (29.5%) went to a company carrying an imaging-manufacturer name — the national subsidiaries of the imaging brands.
• 1,396 awards (70.5%) went to companies with no imaging-manufacturer name attached.
• Widen the definition to count every equipment manufacturer, not only the imaging ones, and the split becomes 730 (36.9%) against 1,251 (63.1%).
• Either way, those awards are spread across 991 distinct winner names.

991 distinct names across 1,981 awards is an unusually flat market. The most frequent single winner in the entire named record appears 34 times out of nearly two thousand awards. There is no dominant national servicer, and no OEM sweep.

The independent tier has names, and they are mid-sized regional specialists rather than global brands: Medist Imaging & P.O.C. (30 awards, plus nine more under a second spelling of its legal name), Power Medical (28), LIAMED (25), Deltamed (18), Niconsulting Technics (13), Silvexim Star, Rafi Medical and Mediserv (10 each), and Varinak Europe, Labo Shop and Helios Dental (9 each) ¹.

On the OEM side, the frequent winners are exactly who you would expect — Siemens Healthcare's Spanish and Czech entities, GE Medical Systems Romania and Poland, Philips Romania and Philips Ibérica ². That is normal and often appropriate: some equipment, some warranty positions and some contractual structures point to the manufacturer.

Three cautions before anyone over-reads this.

It describes the named third of the record. Two-thirds of service awards publish no winner. If unnamed awards skew systematically toward one type of supplier, this picture shifts. We cannot test that, so we state it.

The split moves with the definition. Counting only the imaging brands as manufacturers gives 70.5% non-manufacturer; counting every equipment manufacturer — radiotherapy, dental, laboratory and the rest — gives 63.1%. We report both because a headline that moves seven points on a defensible definition change does not deserve to be quoted as a single number. The direction is what survives either test.

"No manufacturer name attached" is not the same as "independent". Some winners will be authorised dealers, some are distributors reselling manufacturer service, and some are general facilities contractors. The name string tells you what it says and no more.

This is not a claim that OEM service is worse. It is a structural observation about who holds the contracts. The commercially relevant point is narrower and more useful: a company without a manufacturer's badge is not disqualified from this market — it is the majority of it.

6. Where the work is

Takeaway: European countries fall into two procurement cultures, and they call for opposite strategies. Portugal tenders 87 service contracts per ultrasound-equipment contract; Hungary tenders 1.2.

Source: EU TED contract-award notices — Rongtao Medical analysis, July 2026 ¹.

Two patterns, and they are not the same opportunity.

The service-heavy markets — Portugal, Spain, Italy, France, Latvia — put maintenance out to public tender routinely and buy ultrasound equipment through channels that surface far less often in this record. For a company selling service capability, these are the markets where the work is visible, biddable and recurring. Note that Portugal's 87× is a ratio built on a very small equipment denominator (two awards), so read it as "service is what gets tendered here", not as a precise multiple.

The equipment-heavy markets — Poland above all, with 627 of the 1,496 ultrasound-equipment awards in the entire record, plus Hungary and Lithuania — run large, visible equipment procurements. For a parts and repair business, the opportunity there is downstream and slower: those machines are entering fleets now, and they will need parts and board-level support in five to eight years.

Romania sits interestingly in the middle, with high volume on both sides — 146 ultrasound-equipment awards and 1,255 service awards — which is why so many of the named winners in this record are Romanian entities.

7. What the buyer is actually buying

Takeaway: a service contract is sold on two numbers — uptime and response time — and the way uptime is defined is where most of the argument lives.

The award notices record what was committed and to whom. They do not record what the contract obliges the winner to do. For that, the industry's own literature on imaging service agreements is more useful than the procurement record, and it converges on a small set of terms ¹⁰:

• Uptime. OEM service agreements on imaging equipment typically guarantee at least 90%; competing vendors commonly offer 95–98%.
• The definition beneath the number. Uptime is usually calculated on a 24/7 basis rather than over the customer's actual operating hours. A 95% guarantee measured across the whole week is a materially weaker promise than 95% measured across the hours a department is actually scanning — the same percentage, a different obligation. Buyers are advised to insist on the second.
• Response time. A guaranteed response window — four hours is a commonly cited figure — is expected in writing before a purchase order is issued.
• What gets scored. Response time, achieved uptime and service-completion rates are the fields that populate the vendor scorecards procurement teams use to evaluate and re-award contracts.

For a company deciding whether to bid, this is the part that converts capability into a commitment. Multi-brand coverage and parts depth are what let you do the work; an uptime percentage and a response window are what you have to sign. And the two are connected in a specific way: a response-time promise is only as good as the parts you can reach within it. A four-hour response that ends in "the board is on back-order for three weeks" fails the contract on the uptime clause, not the response clause.

That is the mechanism by which parts availability stops being a logistics detail and becomes the commercial risk in a service contract.

8. What it costs to do it yourself

Takeaway: the recognised planning benchmark is that a maintenance programme costs 5–10% of the acquisition value of the equipment it maintains, every year — and WHO is explicit that the low end is only reachable with substantial resources and years of practice.

Any build-versus-partner decision needs a number for "build". The most authoritative one comes from WHO's guidance for health technology management, which defines a cost-of-service ratio: the total annual cost of operating a maintenance programme divided by the acquisition value of the equipment inventory it covers. In the United States that ratio runs between 5% and 10% — and WHO adds the qualification that matters most to anyone planning a new operation, that this range is achievable "only when substantial supporting resources are available, and only after an extended period of performance improvement" ⁷.

Read that against the contract values in section 4. A programme covering an inventory worth €5m carries an annual cost somewhere around €250,000–€500,000 on the WHO ratio. A regional company winning a €455,000 multi-year service award is not being paid to run a mature maintenance programme; it is being paid to deliver a defined scope against defined terms, with the fixed costs of capability spread across every contract it holds.

The staffing side has benchmarks too, and they explain why capability is hard to build at small scale: roughly one technician per 1,000 medical devices, or about 2.6 full-time equivalents per 100 beds; an IFMBE and WHO survey put the total worldwide population of biomedical engineers and technicians at approximately one million across 79 countries ⁸.

Trade coverage of imaging service economics adds a comparative frame that should be read as an industry estimate rather than measured data: annual maintenance is commonly put at 10–15% of equipment cost for an OEM contract, 5–8% for an independent provider and 3–4% for in-house work, with one analysis finding a 57% spread between the lowest and highest full-service quote on the same asset ⁹. Our earlier report on multi-brand fleets covered the savings side of that comparison in detail; the reason to repeat it here is narrower. The spread between models is smaller than the spread between quotes for the same model. Which service model you choose matters less than how well you specify and price the contract — and specifying it well requires knowing what the work actually costs.

9. The European rulebook for an independent servicer

Takeaway: the EU does not regulate third-party servicers as such. It regulates the moment your part or your refurbishment changes what the device is — and the line is drawn in two paragraphs of Article 23.

Our first report in this series covered the American fight over servicing and remanufacturing. The European position is different, considerably shorter, and has not appeared in this series before. It sits in Article 23 of the Medical Device Regulation, headed Parts and components ¹¹.

"Any natural or legal person who makes available on the market an item specifically intended to replace an identical or similar integral part or component of a device that is defective or worn in order to maintain or restore the function of the device without changing its performance or safety characteristics or its intended purpose, shall ensure that the item does not adversely affect the safety and performance of the device."

And the second paragraph, which draws the line:

"An item that is intended specifically to replace a part or component of a device and that significantly changes the performance or safety characteristics or the intended purpose of the device shall be considered to be a device and shall meet the requirements laid down in this Regulation."

Third-party service and maintenance providers are not otherwise singled out in the Regulation. What matters is a second definition: a manufacturer under Article 2 is a person who "manufactures or fully refurbishes a device … and markets that device under its name or trademark". Fully refurbishing a device and putting your own name on it makes you its manufacturer, with every obligation that follows ¹².

Put the two together and the operative boundary for a parts-and-repair business supplying European service work is clear enough to write into a procedure:

• Restore function without changing performance, safety characteristics or intended purpose, and you are supplying a part — with a duty to ensure it does not adversely affect the device's safety and performance.
• Significantly change any of those three, and the item you supplied is itself a device that must meet the Regulation in full.
• Fully refurbish and re-badge, and you are the manufacturer.

That first bullet is not a light obligation, and it is worth reading it the way a public buyer's legal team will. "Shall ensure that the item does not adversely affect the safety and performance of the device" is an outcome duty on the supplier of the part. It is satisfied by evidence — specification control, incoming inspection, traceability and a test that demonstrates the restored device performs as it should — not by an assurance. Which is the same conclusion the recall record pointed to in our report on parts provenance, arriving this time from the statute rather than from the failures.

10. What the record cannot tell you

Takeaway: this dataset is unusually good at some questions and useless for others. Here is the honest boundary.

It cannot tell you what a machine costs. Every value in TED is a whole-contract value. It may cover several years, several lots, several hospitals or an entire equipment category. Dividing by anything to get a unit price would be inventing a number. If you came here for "what does an ultrasound system cost", this dataset does not answer that, and neither does the trade data — the ultrasound customs record we hold reports no country breakdown and no usable quantity unit, so it cannot produce a per-unit price either ³.

It cannot isolate ultrasound service. The repair CPVs cover medical equipment generally. Any reader who takes the €9.85bn as an ultrasound-service figure has misread it, and we would rather say so twice than once.

It cannot see two-thirds of the winners. 4,069 of 6,050 service awards name nobody.

It is not adjusted for inflation or currency. Values are as published, in the record's own terms, across eleven years in which prices moved considerably.

It is European public procurement only, and only above the publication threshold. Private hospital groups, most of the US market, and the large volume of service work procured nationally below the thresholds in section 2 are all outside it. It is a large, clean window — not the world.

It records what was committed, not what was delivered. An award value is the contract's headline figure. Whether the full value was drawn down, whether options were exercised, and whether the service met its uptime terms are not in this record.

What it does support, and what this report claims, is narrower: the relative frequency, relative size, distribution and named-winner structure of publicly tendered medical-equipment service work in Europe. That is enough to answer the question this report set out to answer.

11. What you would need to bid

Takeaway: the small-contract band is winnable, but only by a company that can prove multi-brand coverage, parts availability, a real test step and a quoted turnaround. Most of that can be sourced; one part of it cannot.

The distribution in section 4 is an entry ramp, but it is also a constraint. A €120,000 three-year contract to maintain a mixed imaging fleet will not carry the cost of whole-assembly replacement, long parts lead times, or a technician learning a platform on the customer's machine. What it demands is specific:

One row on that list deserves expanding, because it is the one buyers verify first and the one that cannot be borrowed. ISO 13485:2016 clause 7.5.4 governs servicing activities, and it asks for more than a certificate on the wall: documented servicing procedures with instructions for performing the work, defined qualification requirements for the personnel who perform it, and the records to be kept. It also carries an obligation that catches unprepared organisations — the servicing records must be analysed, including those of the organisation's suppliers, to determine whether the information should be handled as a complaint and fed into the improvement process; if it is decided that something is not a complaint, that decision has to be justified ¹³.

Read that alongside Article 23's duty in section 9 and the shape of a defensible bid becomes concrete. You need a documented procedure, qualified people, a record of what was done, and a loop that turns those records into complaint handling and improvement. A partner can supply the parts, the board-level capability and the test evidence that feeds those records. The system that consumes them has to be yours.

The pattern in that last column is the practical conclusion of the whole report. Two of the requirements — the quality system and the local relationship — have to be yours. Most of the rest is capability that can be bought as a service, which is precisely why 991 different companies can hold contracts in this market without any of them owning a global service infrastructure.

12. Where Rongtao fits

Takeaway: Rongtao Medical is the capability layer behind the column marked "partner".

Rongtao Medical is an independent medical imaging service company founded in 2013 in Guangzhou, serving partners in 140+ countries and regions from a 3,000 m² technical and warehousing centre, with 35+ senior engineers and hundreds of ultrasound testing platforms ⁴.

For a company assembling a service bid, four of those facts are the ones that map onto the checklist above:

• 3,000+ parts SKUs, plus complete systems and probes — the multi-brand depth line.
• Board-level repair and probe solutions — the capability that keeps a small contract profitable.
• A 48-hour real-machine test before shipment, with photos or video where applicable — the evidence a public buyer's acceptance step wants.
• A 5–8 business-day standard turnaround, typical 90-day warranty, and shipping via DHL, FedEx or UPS from a bonded zone with customs paperwork — numbers you can quote in a tender document.

Rongtao holds ISO 13485:2016 and ISO 9001:2015 certification, and covers GE Healthcare, Philips, Siemens, Hitachi, Mindray, Samsung Medison, Toshiba (Canon), Aloka and Biosound Esaote equipment on a compatibility and service basis only — Rongtao is an independent service provider and is not affiliated with, endorsed by, or authorised by any of those manufacturers ⁴.

If you are looking at the numbers in this report and asking whether service is a business you could enter, the useful next conversation is a specific one: which fleet, which brands, which turnaround you would need to promise, and which of the seven rows above you would build rather than source.