← Cross-cutsCross-cut · Quality compounders

The quiet compounders

Seven Teardown studies cover Europe’s “quality compounders” — Halma, Judges Scientific, Spectris, Rational, Geberit, Nemetschek and GEA. None is a household name. Each dominates a dull, narrow niche — combi ovens, concealed cisterns, gas detectors, lab instruments — and compounds quietly for decades. Read together, they reveal exactly what makes the model work, the two ways it’s built, and where every version of it eventually strains.

The model: own a boring niche completely

The common thread is niche dominance most investors never notice. Rational owns ~50% of the global combi-oven market; Geberit is a near-monopoly behind bathroom walls; Halma, Judges and Spectris each own clusters of micro-niches in safety and measurement. Because these niches are small, specialized and unglamorous, incumbents face little big-company competition — and that buys pricing power, which shows up as the fat, durable margins that define the cluster: GEA at 16.5% EBITDA and a remarkable 36.2% ROCE, Geberit at 29.4% EBITDA, Rational at 26.4% EBIT with net cash, Nemetschek at 31.2%. The cross-cut’s first lesson: the most reliable compounding hides in markets too small and too boring to attract a fight.

Two ways to build it: the operator and the acquirer

The seven split into two distinct architectures of the same idea. The focused operators — Rational, Geberit, Nemetschek, GEA — own and deepen one domain, compounding organically through pricing, service attach and conversion of their niche. The serial acquirers — Halma, Judges, Spectris — are the opposite: decentralized roll-ups that buy niche leaders (Judges at ~5× EBIT versus the 12–14× trade buyers pay — the discount itself is the return) and run them hands-off. Halma has bought ~150 businesses since 1983. Reading the operators beside the acquirers shows two routes to the same fortress economics: build the moat, or buy a portfolio of small ones.

The recurring layer is what survives the cycle

Every one of these leans on a recurring annuity to smooth lumpy hardware. Nemetschek is the extreme — 92% recurring software revenue, the “AEC Adobe.” GEA’s ~40% service mix is what lets a maker of lumpy dairy plants hold 36% ROCE; Geberit’s ~60% renovation exposure cushions the new-build cycle; Rational’s ~90% repurchase rate on aftersales is the razor-and-blade engine. The pattern across the cluster: the quality of a compounder is largely the quality of its recurring layer, because that is the part that keeps paying when the capital-equipment or construction cycle turns down — which, for most of these, it has.

Two strains, shared by all: the scale ceiling and the price

Read together, the studies expose where the model breaks. First, the scale ceiling, sharpest for the acquirers: Judges, at just ~£280m, can no longer find deals big enough to move the needle (only £1.9m deployed in 2025), and even Halma’s pace likely slows from a larger base. Compounding by small bolt-ons gets mathematically harder as you grow. Second, the valuation: quality this visible gets fully priced — Halma at ~39× forward (richest in its cohort), Rational ~28×, Geberit ~27×, Nemetschek de-rated from ~62× but still rich. The cluster’s recurring warning is that a wonderful business at a demanding multiple can still be a poor investment — and several here (Judges −65%, Nemetschek −55%) have already shown the air-pocket when growth merely decelerates.

Where they agree — and where they split

All seven share the thesis that niche dominance plus a recurring annuity compounds durably, and all carry a cyclical end-market (labs, construction, food capex) under the quality veneer. They split on the two strains. On durability of the engine: the operators debate saturation (Rational already owns half its market; Geberit faces a weak European cycle), the acquirers debate the scale ceiling and thinning deal supply. On price: the bull reads the multiple as fair for compounding quality; the bear reads it as pricing near-flawless execution — and Spectris is the cautionary endpoint, where a public “quality compounder” trading cheaply was simply taken private by KKR at a 96% premium. The quality of these businesses isn’t the question. Whether the model keeps compounding at scale, and whether you overpay for it, is.

The cluster at a glance

CompanyNicheModelMargin qualityBiggest risk
HalmaLSE: HLMASafety/health/enviro instrumentsSerial acquirer (~150 opcos)21.6% EBIT · 15% ROTIC~39× fwd — richest in cohort
Judges ScientificLON: JDGScientific instruments (~20 cos)Buy at ~5× EBIT, hands-off~19% adj op · 50%+ ROTICScale ceiling at ~£280m · thin deals
SpectrisPrivate (formerly LSE:SXS) — KKR-owned since 4 Dec 2025Precision-measurement instrumentsFocus + bolt-on (now KKR-owned)15.6% adj op (trough)Cyclical + now PE-levered, private
RationalXETRA: RAACombi-steam ovens (~50% share)Premium single-product + aftersales26.4% EBIT · net cashSaturation + China/UNOX value rivals
GeberitSWX: GEBNBehind-the-wall sanitary techInstaller push-pull · ~60% reno29.4% EBITDA · 18.9% RoNSEurope new-build cycle · ~27× multiple
NemetschekETR: NEMBIM / AEC software ('AEC Adobe')Multi-brand SaaS · 92% recurring31.2% EBITDA'AI eats software' · Autodesk scale
GEAETR:G1AFood/dairy/pharma process techLumpy plant + ~40% service16.5% EBITDA · 36.2% ROCEMargin re-rate cyclical or structural?

Figures as of each study’s stated date (2026-06); Spectris is now private (KKR, FY2024 trough figures). Margins, returns and multiples are on mixed bases and not strictly comparable. See each teardown for sourcing and the full weighing.

The seven studies — and the question each turns on

Halma plcLSE: HLMACan a patient, decentralised bolt-on acquirer keep compounding at this pace from a larger base — and is the AI-photonics surge now driving it a structural tailwind or a single-customer concentration risk justifying a ~39x multiple?Read the full weighing →Judges Scientific (JDG)LON: JDGCan the small-acquisition compounding model keep working as Judges scales, or have the law of large numbers, thin deal supply and cyclical demand now capped it?Read the full weighing →Spectris plcPrivate (formerly LSE:SXS) — KKR-owned since 4 Dec 2025Did the 96.3% takeover premium prove Spectris was a mispriced quality compounder, or simply show what cheap-UK listing plus PE leverage will pay for a cyclical industrial?Read the full weighing →Rational AGXETRA: RAAHow much runway is left for a company that already owns ~50% of its niche — and is ~28x earnings paying for runway narrower than it looks?Read the full weighing →Geberit AGSWX: GEBNCan the installer-driven behind-the-wall moat and ~60% renovation annuity defend a ~29% margin and a premium ~27x multiple through a weak European building cycle — or does the price already price the quality?Read the full weighing →Nemetschek SEETR: NEMIs a sticky, recurring-revenue AEC compounder over-sold after a >50% de-rating — or was its premium always priced for perfection against AI-disruption risk, Autodesk's scale and a cyclical construction market?Read the full weighing →GEA Group AGETR:G1AIs GEA's margin re-rating to 16.5% EBITDA / 36.2% ROCE a durable structural reset that compounds toward Mission 30, or a well-executed cyclical peak on flat revenue with the hardest gains still ahead?Read the full weighing →

This is the kind of reading the Desk does for you

A cross-cut takes seven teardowns and one playbook and asks what they say together. The Desk does the same for the companies you actually own — your thesis, the peers that move it, and the tripwires that would change your mind.

See the Desk →