The Cost Centre That Out-Earns Your Core Business

Minus 1.8 percent. That's the gross margin at which GE Aerospace sold its equipment in 2025, above all aircraft engines: $12.2 billion in sales, $12.4 billion in cost. That's no accident. It's the business model. Rolls-Royce spells it out in its annual report: prices for original equipment are agreed with the expected aftermarket business in mind, and "sometimes this may result in losses being incurred on OE." The money comes afterwards. GE's services business runs a gross margin of around 45 percent.

A manufacturer somewhere in Europe, quarterly review. Slide seven, service: "Cost per case down six percent, first-time fix stable, hotline headcount slightly reduced." Satisfied nods. Nobody asks what the two million units in the field could still bring in over their lifetime. That number isn't on any slide. It simply doesn't exist in the building.

Two companies, the same asset. One steers it as a business. The other administers it as a cost block.

The misconception: service as a necessary evil

For most manufacturers, after sales is what comes after the real business. The product is sold, the margin booked, now someone just has to deal with the breakdowns. So service runs like any other cost centre: budget, cost per case, headcount per thousand units. The goal is for it to cost as little as possible.

You can see it in the annual reports. Take 28 industrial manufacturers and read what they publish about their service business. 21 report no separate result for it. Four don't even state service revenue. What nobody reports, nobody steers as a business.

Don't get me wrong: keeping costs under control is never a bad thing. It's just that "cost per case" measures exactly the wrong thing. It rewards everyone who gets rid of the customer quickly. And it punishes everyone who sells something in that moment.

The numbers: where service is visible, it earns more

That leaves the seven out of 28 that show service as a segment with its own result. For them, the picture is unambiguous. In all seven cases the service margin beats the new-equipment or capital-equipment margin — by a factor of 1.4 to 5.2, depending on the company.

In 2025 Otis earned a segment margin of 25.1 percent in service and 4.8 percent on new elevators. Service delivered just over 90 percent of segment profit on roughly two thirds of sales. At Konecranes it's 21.8 against 9.4 percent, at Hiab 23.2 against 12.9, at Nordex 18.1 against 10.5. And in 2023 Vestas sold its wind turbines at an operating margin of minus 1.2 percent while its service business earned 21.0 percent in the same year.

This isn't a lucky find in a handful of filings. In 2025 the Boston Consulting Group looked at roughly 100 industrial machinery companies: service carries gross margins twice the 15 to 25 percent typical of new equipment. Spare parts and consumables average 50 percent.

One caveat belongs here because it's true. The evidence comes from capital goods and machinery. No home-appliance manufacturer publishes a service margin. The mechanics are the same; the public number simply doesn't exist.

The theory: demand you've already paid for

Every unit in the field is a customer you've already won. You know the product, you have the design data, the bills of materials, the service history. Almost no acquisition cost. And a customer who, at the moment of failure, has a real problem and is willing to pay.

That moment is the most valuable one in the entire customer relationship. Repair, spare part, upgrade or straight to a replacement purchase: this is where it's decided who gets the next deal.

Aviation has built an entire business model around it. The engine is the entry ticket; the money is made over the following decades. GE writes in its annual report that competitors offer "substantial discounts" in an effort to secure an installed base that establishes aftermarket sales. Anyone who calculates like that knows what an installed base is worth.

The practice: more service doesn't automatically mean more profit

Now the objection, and it's a fair one. If service is that profitable, why doesn't everyone expand it?

Because it often goes wrong. Research has called it the service paradox since 2005: manufacturers invest in new service offerings, costs go up, the expected returns don't show up. In 2008 Andy Neely analysed more than 10,000 companies. Manufacturers that lean heavily on services generate more revenue on average, but lower margins. And a study in the Journal of Marketing shows that firm value only rises once service reaches a critical mass of 20 to 30 percent of sales. Below that it stays flat or even dips slightly.

The good examples aren't set in stone either. Vestas' service margin collapsed to 12.1 percent in 2024. The company is running a recovery plan and exiting service contracts for other manufacturers' turbines because the portfolio is too diverse to scale efficiently.

That isn't counter-evidence. It's the instruction manual. Half-hearted service doesn't pay. Service on other people's machines without structured knowledge eats margin. What pays is service on your own installed base, with your own knowledge — run as a business with its own profit responsibility, not as an appendix to the warranty department.

The insight: not an accounting mistake, a steering mistake

The segment reports are properly prepared under IFRS and US GAAP. The problem sits one level deeper: in what a company steers its service by.

Steer by cost per case, and you optimise service for invisibility. Steer by the revenue a unit brings over its life, and suddenly you see a business: how many units are where, what a service case is worth today, what the moment of failure could be worth. Same data, different question. And in the end, a different company.

That sounds like semantics. It's organisation. A cost centre has a budget. A business has a P&L, a price, a sales motion and someone who's accountable for it.

Why now: the arithmetic tips in two places

In capital goods, service has almost always been the more profitable business. What's new is that two things are tipping at once that used to make access to the installed base expensive.

First, diagnosis is getting cheap. The expensive step in every service case was always the same: finding out what's broken. It needed an expert, again, for every single case. That step can now be automated at a fraction of its former cost. A study in the Quarterly Journal of Economics covering more than 5,000 support agents measured 15 percent more issues resolved per hour with AI assistance, with the biggest effect among the least experienced. That's the lever for every manufacturer whose service depends on partners and hotlines who don't know the machine inside out. How this new economics tips entire business models is the subject of What Didn't Pencil Yesterday Pencils Today.

Second, regulation makes repair mandatory. The EU repair directive 2024/1799 has applied since 31 July 2026. Manufacturers must repair washing machines, dishwashers, fridges, dryers, vacuum cleaners, displays, smartphones and further product groups beyond the legal guarantee when the consumer asks — free of charge or at a reasonable price. Since 2021, washing-machine makers have already had to keep spare parts available for ten years and deliver them within 15 working days.

Treat that as an obligation and you pay for it. Treat it as a channel and you're with the customer at exactly the most valuable moment: when the machine has stopped and the decision falls between repair and replacement.

Two guardrails before anyone takes a wrong turn

Cutting costs isn't making revenue. AI lowers the cost per case. That makes service leaner, but it stays a cost centre, just a cheaper one. Revenue only appears where someone owns the moment of value: the failure, the spare part, the reorder, the replacement purchase. According to BCG, only 4 percent of manufacturers say they generate recurring revenue from digital services around their equipment. The tools are there. The business model mostly isn't yet.

The asset first, then the automation. An AI business built on unusable data blows up in your face, and it does so in front of your customers. Before any agent diagnoses anything, manuals, bills of materials and service history have to become a queryable model of your installed base. That's the unspectacular part. And the most important one.

What changes: the door is open for others too

There's an effect manufacturers don't like to talk about. The repair directive doesn't only oblige them to repair. It also prohibits them from obstructing independent repairers who use compatible, second-hand or 3D-printed parts.

In other words: the most valuable moment in the customer relationship is being opened up to third parties. Multi-brand service providers, platforms, repair networks. Anyone who runs service as a cost centre has neither a strategy nor a price for that competition. They'll watch others earn money on their installed base.

The zoom-out: Europe's second source of profit is already in the field

European industry has millions of machines, plants, vehicles and appliances in the field. Where new-equipment business is under price pressure, growth isn't in the next unit. It's in the units already out there.

Aviation shows how far you can take this. The elevator industry shows how reliably it carries. For the rest of industry, the question isn't whether there's money in its installed base. The question is who lifts it: the manufacturer itself, or someone who wedges themselves between the manufacturer and its customers.

The principle: build service as a business, don't administer it as a cost centre

What would a system look like that actually taps this source of profit? It would have four properties.

First: steered by lifecycle revenue, not cost per case. The first metric on the slide is what a unit brings in over its life. Cost per case is a constraint, not a goal.

Second: the asset first. Installed base, documentation and service history become a usable data asset before anyone builds an app. The defensible core is the knowledge about your own machines, not the interface on top.

Third: a P&L of its own. A revenue model can't be run inside the warranty P&L. It needs its own profit responsibility, its own pricing, its own sales — in its own vehicle where necessary, so dealer contracts and the core business stay untouched.

Fourth: start on your own base, with critical mass. Your own machines first, maybe other brands later. And better one service business big enough to carry itself than ten offerings that only generate cost.

Coming full circle

GE sells engines at a loss and knows exactly why. The manufacturer in the quarterly review cuts its cost per case and doesn't know what it's giving away. Both paid for their asset long ago. Only one of them earns from it.

Where manufacturers run service as a segment of its own, it earns the better margin. Most run it as a cost centre.

Sources

• Vestas Wind Systems A/S, "Annual Report 2023" (2024), Note 1.1, pp. 57–58: Power Solutions −1.2%, Service 21.0% segment EBIT margin before special items — mb.cision.com
• Vestas Wind Systems A/S, "Annual Report 2025" (2026), pp. 27, 34–35: Service 12.1% (2024) and 16.6% (2025), service recovery plan, exit from multibrand contracts — vestas.com
• Otis Worldwide Corporation, "Form 10-K, Fiscal Year 2025" (2026), Segment Review, p. 34: Service 25.1% vs. New Equipment 4.8% segment operating margin — sec.gov
• Konecranes Plc, "Financial Statement Release 2025" (2026): Industrial Service 21.8% vs. Industrial Equipment 9.4% comparable EBITA margin — konecranes.com
• Hiab Corporation, "Financial Statements Review 2025" (2026): Services 23.2% vs. Equipment 12.9% comparable operating margin — hiabgroup.com
• Kalmar Corporation, "Financial Statements Review 2025" (2026): Services 17.6% vs. Equipment 13.0% comparable operating margin — kalmarglobal.com
• Nordex SE, "Annual Integrated Report 2025" (2026): Service 18.1% vs. Projects 10.5% segment EBIT margin — nordex-online.com
• Valmet Oyj, "Financial Statements 2024" (2025), p. 6: Services 17.4% vs. Process Technologies 3.6% comparable EBITA margin — valmet.com
• GE Aerospace, "Form 10-K, Fiscal Year 2025" (2026), Statement of Operations, p. 35: equipment at a gross loss, services around 45% gross margin (own calculation from sales and cost of sales) — sec.gov
• Rolls-Royce Holdings plc, "Annual Report 2025" (2026), Note 1: original equipment priced against the aftermarket, "sometimes … losses being incurred on OE" — rolls-royce.com
• Boston Consulting Group, "Aftermarket Services Drive Growth and Higher Margins for Industrial Manufacturers" (2025): service gross margins twice the 15–25% on new equipment, only 4% with recurring digital revenue — bcg.com
• Brynjolfsson, Li, Raymond, "Generative AI at Work", Quarterly Journal of Economics 140(2) (2025): +15% issues resolved per hour, largest effect for the least experienced — doi.org
• European Parliament and Council, "Directive (EU) 2024/1799 on common rules promoting the repair of goods" (2024): manufacturers' repair obligation for Annex II product groups, applies from 31 July 2026 — eur-lex.europa.eu
• European Commission, "Regulation (EU) 2019/2023, ecodesign requirements for household washing machines" (2019): spare parts for 10 years, delivery within 15 working days — eur-lex.europa.eu
• Gebauer, Fleisch, Friedli, "Overcoming the Service Paradox in Manufacturing Companies", European Management Journal 23(1) (2005) — doi.org
• Neely, "Exploring the Financial Consequences of the Servitization of Manufacturing", Operations Management Research 1(2) (2008) — doi.org
• Fang, Palmatier, Steenkamp, "Effect of Service Transition Strategies on Firm Value", Journal of Marketing 72(5) (2008): value contribution only above a 20–30% service share — doi.org
• Annual reports of 28 industrial manufacturers (fiscal year 2025): 7 report a service segment result, 21 do not — analysis for this article