UK structural steel: the project giants eat the losses, the product-makers keep the margins

About 184 UK metal-structures manufacturers publish a full profit-and-loss — the firms that fabricate steel frames, bridges, facades, lintels, balconies and mezzanine floors — booking just under £7bn of combined turnover. The map has one clear moral: the two biggest losses in it belong to the two most contract-exposed names. Severfield, the country’s largest structural steelwork contractor, turned £450.9M of revenue into a £17.5M pre-tax loss; McMullen Facades lost £15.8M on £120.8M (a 16-month filing — the latest instalment of a multi-year loss streak). Meanwhile the makers of standardised steel products — Sapphire Balconies (23.8% margin with 74 staff), Leviat (15.0% at the operating line), Keystone Lintels (8.4%) — quietly bank the sector’s best returns at a fraction of the size. In fixed-price steel contracting, scale doesn’t dilute project risk; it concentrates it. Figures are approximate — verify against a company’s own accounts before relying on any single number.

Read the names before the numbers

Two things distort this map, and both sit right at the top of the turnover ranking.

First, about £0.9bn of the headline turnover isn’t building steel at all. Martin-Baker (£382.7M, £89.4M profit) makes aircraft ejection seats; Killinchy Aerospace Holdings (£403.4M) is its parent, so the same ejection-seat business appears in the ranking twice; AETC (£110.3M, £23.9M) makes aerospace engine components. They are classified as metal-structures makers, and technically an ejection seat is a metal structure — but their 22–25% margins are aerospace economics and say nothing about steelwork. The same goes for Vestas-Celtic Wind Technology (£339.7M — the Danish wind-turbine group’s UK sales and installation arm) and Easat Radar Systems (radar antenna structures). We leave them in the tables, labelled, and out of the conclusions.

Second, turnover here moves with the steel price, not just the order book. A fabricator’s selling price is substantially raw material passed through to the client, so a year of steel-price inflation can grow “revenue” with no extra tonnes through the shop — and a year of falling prices can shrink it while the cranes stay busy. Read every turnover-growth figure in this report with that in mind, and treat margin-on-turnover as understating the margin on actual value-added work.

The buyers of what this industry makes have maps of their own: the main contractors who order the steel frames are in our building contractors report, and the infrastructure clients for bridge and gantry work in civil engineering.

The giants

Strip out the aerospace and wind names and the genuine head of the trade is thin: Severfield, William Hare and then a long tier of £25–100M workshops. And the leader is the one bleeding. Severfield’s loss is not a demand story — revenue slipped just 3% and it added staff — it is what happens when fixed-price project risk lands: the accounts carry heavy remediation and contract write-down charges against past bridge and frame work, booked in one year against margins that were thin to begin with. McMullen Facades tells the same story one rung down, with a longer fuse: its £15.8M loss covers a 16-month period to April 2025 after a year-end change (annualised turnover is roughly £91M), and it follows a £33.8M loss on £112.1M in the twelve months before that. Facade packages are the most dispute-prone corner of construction contracting, and McMullen has now lost close to £50M across 28 months — a multi-year bleed that is narrowing, not a single bad year, and still not a story about vanished demand. Contrast William Hare — the second-biggest pure steelwork contractor grew 35% (some of that will be steel price, some data-centre and industrial workload) and kept a 5.9% margin — proof the model can work, but the spread between +£14.8M and −£17.5M at the same trade is the risk premium in one picture. Barclay & Mathieson’s loss is different economics again: it’s a stockholder, whose results swing with the value of the steel sitting in its yards; and Hy-Ten’s 1.3% margin on reinforcing steel, with sales down 12%, is what commodity distribution looks like at the bottom of a price cycle.

The shape of the market

The engine room of this trade is the £5–25M regional fabricator: 114 of the 184 companies, 82% of them profitable. This is a healthy market of substantial workshops — the struggling tier is tiny (only 10 companies under £1M publish full numbers, and just one of those makes money — though most one-shed fabricators don’t publish figures at all). Profitability dips in the £25–100M band (71%) — the tier where firms graduate from the workshop to taking full contractual responsibility for big fixed-price packages.

Products beat projects

Rank the well-run mid-market — profitable operators between £5M and £100M at a 5%+ margin — and a pattern jumps out: the best margins belong to companies that sell a repeatable product, not a bespoke project. Sapphire Balconies makes 23.8% manufacturing offsite balconies with just 74 staff — over £700k of revenue per employee, a design-and-assemble model rather than a steel-bashing one. Leviat sells engineered fixings and connectors into construction worldwide at a 15.0% operating margin — its filed 18.9% pre-tax figure includes £3.9M of interest on group balances with its parent, CRH — though its sales fell 9.3% on the year. Keystone Lintels (8.4%) makes steel lintels by the lorry-load; Davicon mezzanine floors (9.2%); First Fence fencing (9.6%); CST Industries UK storage tanks (12.5%). A product priced per unit lets the maker keep the productivity gain; a project priced per contract hands the client the downside protection and keeps the risk.

The project side isn’t hopeless — it’s just harder. Billington Structures, a structural steelwork contractor a fifth of Severfield’s size, heads the table at 8.8% even as revenue fell back from £118.9M to £97.7M — margin discipline through a down year, and the strongest evidence in the map that the contractor model rewards discipline at a scale where one bad job can’t sink the year.

†Leviat’s 18.9% is a pre-tax margin that includes £3.9M of interest receivable on group balances (it is a wholly owned CRH plc subsidiary); its trading margin at the operating line is 15.0%.

…and 8 more between £25M and £50M, including tank-builder CST Industries UK (12.5%, growing), site-security manufacturer ZND UK (13.1%) and air-handling maker Dalair (13.0%).

Growth, read with care

Remember the pass-through problem: in this trade a +30% revenue year can be a steel-price story. The genuine signal is growth backed by hiring and margin.

Easat Radar Systems’ +249% is lumpy contract revenue on radar structures — flat headcount, 1% margin, not a trend. Martifer UK, the Portuguese metallic-construction group’s UK arm, more than doubled staff to… 28, on £29.2M of revenue — over £1M per employee, which says the labour is subcontracted and the “growth” is project throughput, not a workforce being built. The believable stories are further down: CST Industries UK (+61% with staff up 10% at a 12.5% margin, riding tank and silo demand), Greig Engineering (+57% at a remarkable 20.9%), Davicon (+82% on the warehouse-mezzanine cycle) and Seatrium Offshore Renewable Services (+57%, and its first profit after years of losses — the offshore-wind cycle reaching UK yards just in time). Datalec Power Installations, up in the giants table (+39%, staff +17%), is the same signal at scale: data-centre buildout is currently the strongest single demand stream flowing into this industry.

Market structure: fragmented once you delete the ejection seats

On paper the top five hold 27% of the mapped turnover — modest already. But three of those five are Martin-Baker (twice, via its parent) and Vestas’s turbine arm. Among companies that actually make building and infrastructure steel, no one holds more than about 7% of visible turnover — even the market leader. This is a genuinely fragmented trade of regional workshops, which is exactly what you’d expect where transport costs are high, work is won job-by-job, and scale mostly buys you bigger contract risk.

An old trade, still mostly family-held

Nearly 40% of these companies — 73 of 184 — predate 1990, and the ownership split leans individual (92 individual-owned vs 87 corporate-owned). Only about 10% carry a Holdings/Group/Bidco-style name, the fingerprint of a buyout or planned exit — low by the standards of the markets we map. Private equity has largely left this trade alone, and the giants table explains why: lumpy fixed-price project risk is a poor match for leveraged balance sheets. The exceptions cluster on the product side — Ross-Shire Engineering sits under a Bidco, and Leviat is the UK arm of a global building-products group — which is consistent with the whole report: capital buys the product-makers, not the project-takers.

What the map shows

1. Project risk lands at the top. The market leader (Severfield, −£17.5M) and a major facade contractor (McMullen, −£15.8M over a 16-month period, on top of −£33.8M the year before) took the map’s two biggest losses — write-downs on fixed-price contracts, not vanished demand.
2. Products beat projects. Balconies, lintels, fixings, tanks and mezzanines earn 8–24%; bespoke steelwork contracting earns low single digits in a good year and eats the losses in a bad one.
3. Turnover is part steel price. Selling prices pass raw material through to the client, so revenue growth and decline both overstate what’s happening to actual workload.
4. The £5–25M workshop is the healthy heart — 114 companies, 82% profitable — and profitability dips, not rises, as firms graduate into big-ticket contracting.
5. Data centres and offshore wind are the live demand streams — Datalec +39%, Seatrium +57%, CST +61% with hiring behind them.
6. A tenth of the map is mislabelled aerospace. Martin-Baker’s ejection seats and AETC’s engine parts wear a metal-structures label; their margins belong to a different industry and are excluded from every conclusion above.

Methodology and caveats

This covers only the 184 UK metal-structures manufacturers that publish a full profit-and-loss; the long tail of small fabrication shops files abridged accounts with no figures and doesn’t appear. The £6.95bn combined turnover overstates the distinct total: roughly £0.5bn is double-counting where a parent and its subsidiary both publish numbers (Killinchy consolidates Martin-Baker; Severfield consolidates its Northern Ireland arm), and around £0.85–0.9bn more is aerospace and wind-energy business carrying a metal-structures label (Killinchy’s ejection seats, AETC’s engine parts, Vestas’s turbine arm, Easat’s radar structures — counting each group once). Turnover in this trade embeds volatile steel prices, so year-on-year revenue moves are part price, part volume; large one-off losses may be contract write-downs, remediation provisions or disputes rather than ordinary trading; margins are never compared across different business models here (product manufacture, fixed-price contracting, steel stockholding and distribution are different economics). Extreme proportional outliers are excluded from the charts. Figures are approximate and business labels are directional — verify any specific figure against the company’s own accounts. This is analysis, not financial advice.