U.S. Foundries (NAICS 3315): A Rollup Investor Primer

1. Overview

North American Industry Classification System (NAICS) code 3315 — Foundries is the U.S. industry group that takes solid metal, melts it, and pours it into molds to make near-net-shape parts. It splits cleanly by metal chemistry into two child industries: 33151 Ferrous Metal Foundries (iron- and steel-based castings) and 33152 Nonferrous Metal Foundries (aluminum, copper, zinc, magnesium, nickel and superalloy castings).[1] Between them they make the cast parts buried inside almost every physical thing that moves or holds pressure — vehicles, aircraft engines, farm and construction machinery, pumps and valves, water and sewer systems, rail cars, mining equipment, power turbines, and defense platforms.[3][4]

The two children run the same core physics — melt, pour, cool, clean, heat-treat, inspect, machine — but they are different businesses with different customers, different owners, and different ways in for an outside investor. The value of this rollup is the contrast between them, and the one place where they overlap. Ferrous is the larger, heavier, more concentrated half built around fewer big plants; nonferrous is the more fragmented, lighter, more auto-exposed half built from many small plants. Yet the clearest public-market exposure in the entire group does not sit in either half's bulk — it sits in the narrow precision investment-casting corner that spans both: cast steel parts on the ferrous side and superalloy turbine blades on the nonferrous side, sold into aerospace and power generation.[3][4]

Three threads run through the whole group. First, there is no clean U.S.-listed pure play in foundry casting. Listed access did widen in 2026 — a precision caster (Doncasters/DPC Holdings) listed on the New York Stock Exchange on June 25, 2026, and a small holding company (Air T) carries a direct 20.1% minority stake in an iron foundry — but both are partial, and exposure otherwise runs through diversified manufacturers, captive in-house foundries, or upstream alloy suppliers.[3][4] Second, most of the actual industry is private — family-owned, employee-owned, and private-equity (PE) held. Third, and newly quantified in the revised child pages, profitability across this group tracks qualification difficulty, not tonnage: the reported margins of the automotive- and municipal-facing bulk sit in the low single digits, while the aerospace precision corner reports margins in the twenties and thirties.[3][4] This is a capital-intensive, cyclical, operationally unforgiving field where capacity utilization, casting yield, energy cost, and defect rates decide profits more than headline revenue does.

2. What's inside — the two children and how they differ

The group divides on one variable: is the metal iron/steel (ferrous) or not (nonferrous)? That single split drives almost everything else — plant size, tonnage, customer mix, concentration, energy intensity, and ownership. Shares below are shown three ways because they tell different stories, and all three reconcile exactly to the group total (see Section 3).

Contrast table

How the two halves differ economically

Both children buy or make metal, run melt furnaces, and live or die on how full those furnaces stay. The differences are in emphasis:

• Ferrous is a tonnage-and-conversion business. Iron foundries — the largest single piece of the whole group — think in tons and price-per-pound, are freight-sensitive (heavy parts don't ship far economically), and are regional as a result. They are also the most energy-hungry corner per dollar of output in the group.[3][11] Cast steel adds heavy safety-critical parts (mining, rail, energy) and a small, extraordinarily value-dense precision niche.[3]
• Nonferrous is a lightweighting-and-volume business. Die-casting — the largest child within nonferrous, at 55.7% of its employment — is a press-hours-and-cycle-time business tied hard to automotive volumes, making it the most cyclical corner of the group. Aluminum sand/permanent-mold work carries more machining content; the "other nonferrous" superalloy tail is the highest-specification, longest-qualified work in the group and the only corner with a rising published price index.[4]
• The plants look different. Ferrous runs roughly twice the revenue per plant and nearly twice the headcount per plant of nonferrous.[2] Nonferrous is more numerous and smaller — 811 firms operating just 819 plants means almost every nonferrous firm runs a single site.[4]
• Even Washington sizes them differently. Small Business Administration employee ceilings across the group's six grandchildren range from 500 (steel except investment; aluminum except die-casting; other nonferrous) to 700 (die-casting), 1,000 (iron) and 1,050 (steel investment) — a more-than-twofold spread inside one four-digit code, with practical consequences for set-aside eligibility and lending.[3][4][10]

The single most useful takeaway: the two halves do not rise and fall together. A soft farm-equipment or municipal-budget year hurts iron; an auto downturn hits nonferrous die-casting hardest; and an aerospace ramp lifts the precision corner of both. The two published output-price series now make that visible — iron prices up ~2.6% year over year while other-nonferrous primary products rose ~11% in 2025 and 9% again in five months of 2026.[3][4] That partial de-correlation gives the aggregated group some natural diversification that neither child has on its own.

3. Size — the rollup figures

Our ground-truth federal statistics for NAICS 3315 as a whole:

Average pay works out to roughly $67,000 per employee across the group — ferrous a little higher (~$69,500) and nonferrous a little lower (~$64,600), consistent with ferrous's heavier, more skilled steel work.[2][3][4] Every metric here is present in our ground-truth file; none is suppressed. Receipts, employment, and establishments reconcile to the two children exactly, and payroll reconciles to rounding ($3.97B + $3.243B).[5]

A note on the firm count: the two children report 411 + 811 = 1,222 firms, but the group total is 1,211 — about eleven firms run both ferrous and nonferrous foundries and are therefore counted once at the group level but twice across the children.[2][3][4][5] It is a small reconciliation detail, not a data error.

The group is measured far better than its parts. This four-digit level and its two children publish receipts, firm counts, and concentration ratios; below them, federal revenue data essentially disappears. Of the group's six grandchildren, exactly one — steel investment foundries (331512) — has a current, measured Economic Census receipts line: $3.95 billion across 84 firms, about 20% of ferrous receipts and roughly 12% of the group's total.[3][5] Iron's most recent directly reported benchmark is 2017 ($8.584 billion of shipments); steel-except-investment has only a modeled OSHA regulatory estimate ($4.646 billion, 164 firms, in 2022 dollars); and all three nonferrous grandchildren reported that no publishable six-digit receipts, shipment, firm-count or concentration figure was available at all.[3][4] Different vintages, universes, and methods — these cannot be summed, against each other or against the group total. What the evidence supports is ordering, not addition.

Concentration — and why the group number understates it. For the group as a whole, the four largest firms hold 20.2% of receipts, the top eight 30.4%, the top 20 48.2%, and the top 50 63.7%; the Herfindahl-Hirschman Index (HHI — the sum of squared market shares on a 0–10,000 scale, where regulators generally treat readings below ~1,000 as an unconcentrated market) is just 170.4.[2] That reads as a highly fragmented industry — and in aggregate it is. But note the oddity: the group HHI (170.4) is lower than either child's (ferrous 337.4, nonferrous 177.6).[2][3][4] Aggregating two partly separate product markets spreads share across more firms and mechanically lowers measured concentration, so the group number is the least informative figure about real competitive structure. The concentration lives further down: within ferrous, the steel-investment grandchild has its top four firms controlling roughly 73.5% of receipts, the top eight 79.5%, the top 20 89%, and the top 50 97.9% — up from a CR4 of 59.2% and an HHI of 1,662 across 130 companies in 2002.[3] The catch is that this is the only grandchild in the entire group with published concentration ratios, and even there Census suppressed the HHI.[3][4] Always look one or two levels down before concluding a foundry market is fragmented — and expect the data to run out when you do.

Sources: group figures[2]; ferrous and nonferrous figures from the child primers[3][4].

The base is projected to shrink. The Bureau of Labor Statistics projects employment across NAICS 3315 — this exact group — falling from 106,000 in 2024 to 90,400 in 2034, a 14.7% decline, with real output slipping from $26.9 billion to $24.8 billion.[3][9] (BLS works from its own employment base, slightly different from CBP's 107,273, and its real-output measure is not the same construct or price basis as the Census receipts line above.) That decade view sits alongside a near-term upswing — an American Foundry Society survey projects overall casting sales up 4.2% in 2026, of which 52% is pricing and only 48% tonnage, with 72% of respondents positive and 95% making capital investments.[3][16] The two are not contradictory: a pricing-led cyclical recovery inside a structurally consolidating, automating, shrinking base. Where child data reaches back, the attrition is already visible — steel investment from 130 companies in 2002 to 84 firms today, and steel-except-investment from 248 establishments and 16,831 employees to 141 and 9,928.[3]

Undercount and comparability caveats. These are employer establishment figures. County Business Patterns omits nonemployers (owner-only businesses), but in this furnace-intensive, capital-heavy field that is a minor gap — foundries are not a sole-proprietor or government-dominated trade. The material blind spots are elsewhere: (a) captive foundries run inside vehicle, engine, machinery, marine, and firearm makers are counted under the parent's finished-product industry, not here, so the true physical foundry footprint is larger than 1,335 plants;[1][3][4] (b) the marquee aerospace product — the single-crystal superalloy turbine blade — is nonferrous and sits in NAICS 331529, so the economic weight of the companies that lead precision casting is split across both children rather than showing up in one place;[3][4] (c) small family- and individually-owned bronze and copper foundries at the low end of nonferrous make the very smallest tier easy to under-count;[4] and (d) any long time series for die-casting — the group's largest grandchild by headcount — must respect the 2012 NAICS revision that merged codes 331521 and 331522 into today's 331523, or it will manufacture apparent growth or decline.[4] No suppressed value is stated as if known.

4. Investable universe — where value concentrates across the children

The central rollup insight for investors: public-market value is inversely related to the industry's physical center of gravity. The biggest, highest-tonnage, highest-headcount corners — iron foundries and die-casting — offer almost no clean listed exposure. The value-dense precision corner that straddles both children is where the public proxies cluster, and it is the smallest measured piece of the group.

The crossover: aerospace and turbine precision casting. This is the one place ferrous and nonferrous meet on the public market. Cast steel investment parts (ferrous, NAICS 331512) and superalloy turbine blades (nonferrous, NAICS 331529) are made by many of the same companies:

• Howmet Aerospace (HWM) — the cleanest large-cap precision-casting proxy, spanning cast steel and nonferrous superalloy; ~$8.3 billion of 2025 revenue, roughly 70% from commercial and defense aerospace, with its Engine Products segment posting $4.320 billion of third-party sales and a 33.3% adjusted EBITDA (earnings before interest, taxes, depreciation, and amortization) margin, up from 30.8% in 2024.[3][4]
• Berkshire Hathaway (BRK.B / BRK.A) — owns Precision Castparts (PCC), the global precision-casting leader (~33,000 employees; PCC Structurals, PCC Airfoils, Wyman-Gordon), bought for ~$37.2 billion in 2016, but only as one holding among many. The children disagree on PCC's 2025 revenue: the ferrous primer reports ~$10.4 billion and the nonferrous primer ~$10.8 billion. Treat it as "roughly $10–11 billion" rather than choosing between them.[3][4]
• Doncasters / DPC Holdings (DPC) — listed on the NYSE on June 25, 2026, selling 27,858,585 shares at $33 for approximately $919.3 million before expenses; $837 million of fiscal-2025 revenue, 77% from aerospace and industrial gas turbines. It offers the clearest casting economics of any listed name, but it is primarily a nickel- and cobalt-superalloy producer — nonferrous work — rather than a steel investment foundry, which is why both child primers claim it and neither owns it outright.[3][4]
• ATI (ATI) and Carpenter Technology (CRS) — specialty-alloy suppliers upstream of the foundries; ATI reported $4.587 billion of 2025 sales with precision forgings, castings and components together at 22% of revenue, casting not separately disclosed.[3][4]
• Impro Precision (HKEX: 1286) — an international comparable whose investment-casting segment ran HK$1.91 billion in 2025, 37.5% of the group; global and alloy-mixed, not U.S. exposure.[3][4]

Ferrous-specific public routes are all indirect and diluted:

• Iron: Mueller Water Products (MWA) runs two U.S. iron foundries for valves and hydrants on $1.430 billion of fiscal-2025 sales but reports no standalone foundry results; Air T (AIRT) holds the only direct listed ownership of a merchant iron foundry — a 20.1% equity-method interest in privately held Cadillac Casting on an original investment of $2.8 million, buried inside an aviation-dominated holding company; Caterpillar (CAT) ($67.589 billion of 2025 sales; its Mapleton, Illinois, foundry can melt as much as 1,000 tons of iron per day) and Deere (DE) ($45.684 billion fiscal-2025) run captive iron foundries and are better read as end-market indicators.[3]
• For an actual listed iron-foundry income statement you have to leave the U.S.: UK-listed Castings plc reported £171.746 million of external foundry revenue and just £6.378 million of foundry segment profit for the year to March 2026 — under 4%, and a useful warning about how thin iron economics get when volumes or execution disappoint.[3]
• Heavy steel (except investment): Ampco-Pittsburgh (AP), whose Forged and Cast Engineered Products segment did $292.6 million of 2025 sales, and the UK-listed Weir Group (WEIR), whose ESCO division generated £709 million of 2025 revenue with North America at 56% of orders; neither breaks out U.S. steel-casting revenue.[3]
• The best iron and heavy-steel assets are private: Waupaca Foundry (PE firm Monomoy; 4,000-plus employees, five foundries, 1.4 million tons of capacity), Grede (PE firm Gamut; seven North American facilities), Charlotte Pipe/Neenah, family-owned McWane, employee-owned AMERICAN Cast Iron Pipe and Amsted, WHEMCO/Park, Vulcan Metals (TerraMar), Harrison Steel, employee-owned Eagle Alloy, Metal Technologies.[3]

Nonferrous-specific public routes are thinner still, especially in the largest child:

• Die-casting (331523), the biggest nonferrous child by headcount, has no clean U.S. proxy. Ryobi (TSE: 5851) is the purest operating comparable — die castings were 88.7% of fiscal-2025 sales — but it is a Japanese, globally diversified automotive supplier; Nemak (BMV) is Mexico-listed; Linamar (TSX: LNR) bought Aludyne's North American assets for an initially valued $300 million and buries casting inside a diversified mobility group; Allison Transmission (ALSN) owns Walker Die Casting as a small vertical integration.[4]
• Aluminum (except die-casting): captive exposure via Brunswick (BC), whose Mercury Marine pressurized lost-foam foundry sits inside a $2.177 billion 2025 propulsion segment, and General Motors (GM), which announced an investment exceeding $150 million in 2026 at its Saginaw precision-sand and semi-permanent-mold plant; the private base includes Wisconsin Aluminum Foundry, Ligon, CaneKast, Angstrom, Eck Industries, Bremer, and Amsted.[4]
• Private PE platforms dominate: Consolidated Precision Products (Warburg Pincus/Berkshire Partners), Form Technologies (Ares funds, owning both the excluded die-caster Dynacast and the in-scope precision caster Signicast), Pace Industries (MiddleGround), Gibbs Die Casting (Architect Equity), MetalTek, Winsert (Altus).[4]

Three patterns are visible only from this level. Employee ownership is a recurring structure on both sides of the metal split, not a ferrous quirk — Amsted, Eagle Alloy and AMERICAN Cast Iron Pipe on the iron and steel side, Amsted again on the nonferrous side — which shapes both deal availability and price expectations.[3][4] Company boundaries do not respect the ferrous/nonferrous line at all: Amsted and MetalTek appear in both children's rosters, Howmet spans both, and Form Technologies owns one business inside the group and one excluded from it — so any firm-level screen built on NAICS codes will straddle them.[3][4] And across the whole group, the biggest and best assets are private, and the public tickers give blended, diluted exposure. Corporate segment revenue for any listed name should never be read as NAICS industry revenue — segments fold in foreign plants, adjacent forging and machining, and excluded metals. Valuations and yields appear in Section 10.

5. How the money works

Both children run the same revenue-and-cost architecture, with different accents:

• Revenue ≈ parts (or tons) shipped × negotiated price, plus tooling, machining, coating, heat treatment, and assembly. The most valuable operators ship qualified, installation-ready components, not commodity cast metal. Ferrous iron foundries price in tons and cents-per-pound; die-casters price in press hours and piece prices with annual "price-downs"; precision casters price per part under multi-year long-term agreements (LTAs) — Doncasters covered roughly 70% of 2025 revenue with LTAs.[3][4]
• High fixed cost, high operating leverage. Furnaces, molding and casting lines, and pollution controls cost the same whether busy or idle, so capacity utilization is the master dial — modest volume swings produce large margin swings, up fast and down faster. Iron-and-steel-product capacity utilization (a broad proxy, not a foundry-only measure) ran 72.2% in 2025, implying real slack. Track good castings shipped, not pounds or tons melted.[3][4]
• Metal is largely a pass-through; the foundry keeps a conversion margin. Scrap, pig iron, aluminum, and alloys (nickel, chromium, cobalt) are typically passed to customers via surcharges — aluminum priced off the London Metal Exchange (LME) plus a regional premium — so the plant earns the margin on its melting, molding, and machining. Exposure is also less "primary" than it looks on the nonferrous side: the North American Die Casting Association states that more than 95% of North American aluminum die castings use post-consumer recycled aluminum, which needs roughly 5% of the energy of primary metal.[4] Pass-throughs recover cost with a lag everywhere, which squeezes margins when input prices spike.[3][4]
• Energy is a first-order cost, and the melt decision is a capital decision. Energy intensity varies nearly twofold across the group's measured grandchildren (see §2), and furnace choice follows: AMERICAN Cast Iron Pipe is spending $285 million through 2029 to replace a cupola with four coreless induction furnaces, expecting 25% more melting capacity and a 50% reduction in on-site greenhouse-gas emissions — trading combustion exposure for electricity-price and grid exposure.[3][11]
• Yield and quality are the profit levers. Casting yield (good metal out vs. metal poured), gates/risers/porosity losses, and first-pass scrap/rework rates decide unit economics. A defect found after pouring, heat treatment, and machining destroys nearly all the cost accumulated to that point, not just the metal; on safety-critical aerospace and turbine parts it can mean warranty, a delivery penalty, or permanent disqualification.[3][4]
• Stickiness and switching costs — with the bill attached. Patterns, dies, tooling, and customer qualifications make relationships durable; in aerospace and defense an incumbent's certified, often sole-source position is a genuine moat. The flip side is customer concentration: Doncasters' two largest customers alone were 38% of 2025 sales, and its ten largest were 68–69% (the two child primers report the ten-customer figure one point apart). The moat and the concentration risk are the same fact.[3][4]
• Aftermarket smooths the cycle. Replacement/wear parts are higher-margin and steadier than original-equipment volume — at Weir's ESCO division, aftermarket was 94% of 2025 revenue (a company example, not an industry average).[3]

The margin ladder — the clearest new picture this rollup can give. No audited margin data exists for these NAICS codes themselves, and the available company proxies are measured on incompatible bases — different companies, currencies, periods, segment definitions and profit measures (operating margin vs. EBITDA vs. adjusted EBITDA). None is an industry average, and they cannot be compared decimal-for-decimal. Read the ordering, which runs across both children rather than within either:

• Iron (ferrous): Castings plc's £6.378 million of foundry segment profit on £171.746 million of revenue — under 4%, before other operating income and pension cost.[3]
• Die-casting (nonferrous): Ryobi's die-casting segment at a 4.1% operating margin in 2025, guided to 4.4% for 2026; Nemak at ~6% operating and 14% EBITDA in the second quarter of 2025; GF Casting Solutions' automotive business at $91 million of adjusted EBITDA on $707 million of 2024 revenue (about 12.9%) before its sale to Nemak.[4]
• Heavy steel (ferrous): a modeled ~4.7% pre-tax profit rate from an OSHA regulatory analysis — a screening estimate, not observed EBITDA. Actual results can be far worse: Ampco-Pittsburgh's casting-related segment posted a $44.7 million operating loss in 2025 on $292.6 million of sales, most of which was a $41.4 million deconsolidation charge plus $10.8 million of exit charges — evidence that restructuring can overwhelm revenue scale, not a run-rate loss.[3]
• Precision casting (spanning both children): Doncasters at 23.1% gross margin and 16.5% adjusted EBITDA margin for 2025 — split 18.0% in North American aerospace, 22.0% in European industrial gas turbines, and just 6.5% in transportation turbo wheels — and Howmet's Engine Products at 33.3%.[3][4]

The gap between roughly 4% and roughly 33% is far too wide to be an artifact of definitions. It is the single clearest quantitative expression of this group's central asymmetry — margin rises with qualification difficulty and value density, not with tonnage or headcount — and the same spread reappears inside the precision corner, where Doncasters' turbo-wheel work earns under a third of what its gas-turbine work does.

For private transactions, EBITDA should be normalized for metal pass-through timing, tooling income, and launch costs, and maintenance capital expenditure is frequently higher than book depreciation implies — old furnaces and presses hide real reinvestment needs. Metrics owners and analysts watch across both children: utilization, tons/parts shipped and price, casting yield and scrap/rework, energy and labor cost per unit, on-time delivery, backlog and book-to-bill, LTA/aftermarket coverage, and maintenance capital.[3][4]

6. Demand drivers

Because the two children point at overlapping but differently weighted end markets, the group is diversified in aggregate even though each half is cyclical on its own:

• Vehicles drive both halves but differently: heavy trucks, ag and construction machinery pull iron; light-vehicle lightweighting pulls aluminum die-casting and sand casting. Transportation is the single largest U.S. aluminum end market (an estimated 36% of aluminum consumption in 2025 — an aluminum-wide figure, not a NAICS revenue share). The physics has a quantified payoff: the Department of Energy estimates a 10% vehicle-weight reduction improves fuel economy by 6–8%, and credits aluminum and aluminum-matrix composites with 30–60% mass-reduction potential depending on application.[4]
• Aerospace, defense, and industrial gas turbines anchor the precision corner of both children — cast steel and superalloy alike — and were described as capacity-constrained heading into 2026. Commercial aerospace alone accounted for over 45% of global investment-casting demand in 2024; Howmet's defense-aerospace revenue rose 21% in 2025, its industrial-gas-turbine revenue 25%, and Engine Products sales 16%. Certified cast turbine and engine parts are hard to substitute, giving qualified suppliers recurring aftermarket demand; gas turbines additionally ride data-center-driven electricity demand.[3][4]
• The electric-vehicle (EV) transition is a redistribution, not a simple tailwind. EVs erode internal-combustion-engine (ICE) castings on both sides — iron blocks/heads/manifolds and aluminum blocks/housings — while creating demand for large structural aluminum "gigacastings" (favoring die-casting) and leaving brakes, hubs, suspension, and water infrastructure untouched. The repositioning is measurable: the GF Casting Solutions automotive portfolio Nemak acquired was roughly 80% e-mobility and structure/chassis. Against that, General Motors' Saginaw investment in next-generation engine blocks and heads shows profitable conventional-engine programs persisting. The net effect depends on each foundry's product portfolio, not headline vehicle sales.[3][4]
• Infrastructure and municipal water provide ferrous's steadiest demand — ductile-iron pipe, valves, hydrants, and castings replaced on long cycles and supported by domestic-content rules.[3]
• Mining, rail, steel mills, and energy drive heavy cast steel — crusher and ground-engaging wear parts, mill rolls, couplers, pumps, and valves. The strongest secular argument here is not volume growth but scarcity of qualified domestic capacity, which Defense Department industrial strategy emphasizes.[3]
• Input conditions cut across the group: U.S. aluminum spot-ingot prices rose 39% in 2025 against net import reliance of 60%, while stainless prices eased about 11% as nickel oversupplied — foundries without timely pass-throughs bear the margin risk in either direction.[3][4]
• Lightweighting, localization, and reshoring support domestic casting across the whole group.[3][4]

Near-term optimism sits inside a long-run contraction, and both are in the evidence. The AFS survey's projected 4.2% 2026 sales growth is 52% pricing and only 48% tonnage;[3][16] BLS projects the group's employment down 14.7% by 2034.[3][9] Useful leading indicators: vehicle and ag/construction build rates, aircraft build rates and airline backlogs, defense procurement, mining and steel-mill capital spending, railcar orders, electricity-demand and grid investment, municipal water budgets, LME aluminum and nickel prices, and interest rates (which drive the rate-sensitive vehicle and equipment cycles).[3][4]

7. Regulation

Regulation is largely common across both children, since they run similar high-heat, dusty, metal-melting processes:

• Air emissions. The Environmental Protection Agency (EPA) regulates foundries under National Emission Standards for Hazardous Air Pollutants (NESHAP) in Title 40 of the Code of Federal Regulations (CFR), Part 63 — iron and steel foundries under Subpart EEEEE (major sources) and Subpart ZZZZZ (area sources); aluminum, copper, and other nonferrous foundries under their own nonferrous subparts, with nonferrous area sources under Subpart ZZZZZZ.[3][4][6][7] Controls target particulate matter and hazardous air pollutants — manganese, lead, chromium, nickel, beryllium, cadmium, and organics from melting, molding, and cooling; high-alloy and stainless work raises the chromium and nickel burden. Applicability can turn on charge material and furnace practice rather than NAICS code: nonferrous plants charging secondary scrap may instead fall under the secondary-aluminum NESHAP.[4]
• Worker safety. The Occupational Safety and Health Administration (OSHA) limits respirable crystalline silica from foundry sand to an eight-hour permissible exposure limit (PEL) of 50 micrograms per cubic meter, with a 25 action level.[8] Molten metal, heat, noise, grinding, fume, machine guarding, and hazardous-energy control add obligations; combustible aluminum and magnesium dust add explosion and burn risk specific to nonferrous plants.[3][4]
• Water and waste — with a partial offset. Clean Water Act effluent guidelines cover foundry wastewater — quench, mold-cooling and scrubber streams — under 40 CFR Part 464, with some process segments set at no discharge of process wastewater pollutants.[4][13] Spent sand, baghouse dust, dross, and slag require characterization and disposal or reuse compliance, but the reuse channel is real: EPA estimates roughly 2.6 million tons of foundry sand are beneficially used outside foundries each year, with iron, steel, and aluminum sands accounting for 96% of that volume — so spent sand can be a cost or a modest recovery depending on state rules, contamination, and local end markets.[3][12]
• Trade — protective and costly at the same time. Section 232 metals tariffs cut both ways across the group: as of mid-2026, roughly 50% on covered primary steel, aluminum, and copper articles and 25% on predominantly-metal derivatives, subject to Harmonized Tariff Schedule classification and metal-content tests, with some unfinished cast-steel articles carrying the additional 50% duty.[3][4][14] Product-specific antidumping and countervailing orders add targeted shelter in individual niches. But the U.S. International Trade Commission found that Section 232 and Section 301 tariffs, while reducing imports and raising U.S. production and prices, left U.S. importers bearing nearly the full tariff cost,[3][15] and duties on castings do not stop imported finished machinery that already contains foreign castings. Verify by product rather than assuming uniform protection.

Aerospace/defense adds extra gatekeepers in the precision corner of both children: the AS9100 quality standard, Nadcap (originally the National Aerospace and Defense Contractors Accreditation Program) accreditation for special processes, program-specific OEM and FAA source qualification, ITAR/EAR export controls, and the Defense Federal Acquisition Regulation Supplement (DFARS) specialty-metals clause, which generally requires U.S.-melted specialty metals for defense parts and structurally favors domestic foundries.[3][4]

8. Consolidation

Consolidation has been real and is led by private equity, but it has not concentrated the group — with a group HHI of 170.4 and a top-four share of 20.2%, foundries remain highly fragmented in aggregate even after a decade of deal-making.[2] The action is uneven across and within the children:

• The precision corner is the most consolidated and still consolidating — and now the change is measurable. The steel-investment grandchild went from 130 companies with a CR4 of 59.2% and an HHI of 1,662 in 2002 to 84 firms with a CR4 of 73.5% today (HHI suppressed).[3] Original-equipment makers prefer fewer, fully qualified suppliers; Berkshire's 2016 PCC purchase (~$37.2 billion) was the defining deal; PE roll-ups (CPP, Signicast) gathered smaller shops into platforms, and the closest listed proxy moved from private ownership to the public market via the June 2026 Doncasters IPO. High qualification, capital, and metallurgical barriers keep new entrants scarce.[3][4]
• Ferrous iron and heavy steel consolidate regionally and selectively — and shrink by attrition. Assets pass among industrial groups, families, and PE sponsors (Waupaca to Monomoy in 2024; Grede to Gamut in 2019; Neenah Enterprises to Charlotte Pipe in 2022; TerraMar buying Bradken's North American specialty foundries and machining operations in 2025), but plants are hard to relocate, environmental liabilities are large, and customer qualifications may not transfer, so buyers pursue focused carve-outs rather than sweeping roll-ups. The antitrust ceiling is real: in the 2021 Neenah/US Foundry municipal-castings deal the Department of Justice found the parties were two of only three significant suppliers across 11 states and forced divestiture of rights to more than 500 patterns. Steel-except-investment has meanwhile fallen from 248 establishments and 16,831 employees in 2002 to 141 and 9,928.[3]
• Nonferrous die-casting and aluminum are actively rolled up but stay fragmented. PE platforms (MiddleGround/Pace, Ares/Form Technologies including Dynacast, Architect/Gibbs) and strategic acquirers gathered plants — Linamar bought Aludyne's North American assets for an initially valued $300 million and Nemak completed its acquisition of GF Casting Solutions' automotive business — while aluminum roll-ups broadened process and machining coverage, with Wisconsin Aluminum Foundry adding low-pressure specialist ATEK in 2024 and tooling maker Anderson Global in 2025. Yet single-plant firms still dominate the child.[4]

The consistent pattern across the whole group: buyers pay for customer-qualified capacity, machining depth, and scarce process approvals, not merely cheap furnaces. National scale alone is not a moat; patterns, tooling, environmental permits, skilled labor, rare furnace and lifting capacity, and customer qualifications are the barriers that protect incumbents and deter new plants. Note also that establishment counts are a poor proxy for share — one company can own several plants, while a single foundry can dominate a narrow casting size or qualification.[3][4]

9. Risks

Both children share most risks, differing mainly in which end-market cycle bites:

• Cyclicality plus fixed-cost leverage: vehicle, ag, mining, rail, marine, and aerospace volumes can fall faster than fixed costs, and low utilization sharply compresses margins — sharpest where reported margins already sit in the mid single digits (die-casting's automotive book, iron conversion work). Ampco-Pittsburgh's casting-related segment posted a $44.7 million operating loss in 2025, mostly restructuring and deconsolidation charges rather than a run-rate loss.[3][4] The two halves' cycles are only partly correlated, giving the aggregate some diversification.
• Customer and program concentration — the aggregate hides it. The group is unconcentrated on paper (HHI 170.4) but individual operators are not: Doncasters' two largest customers were 38% of 2025 revenue and its ten largest 68–69%. Losing one qualified vehicle, aircraft, or turbine program can idle dedicated presses, patterns, and machining cells.[3][4]
• Single-plant and sole-source disruption: qualification is both a moat and a chokepoint. Berkshire disclosed that a February 2025 fire at Precision Castparts' Jenkintown facility affected more than 700 sole-sourced parts considered critical by aerospace customers.[4]
• Input and energy volatility: scrap, iron, aluminum, nickel, cobalt, electricity, coke, gas, binders, refractories, and freight can move faster than customer pricing; surcharge pass-throughs lag, and volatility cuts both ways on conversion margin.[3][4]
• Substitution and technology: EV adoption redistributes demand; shifts toward aluminum, forgings, weldments, composites, or fabricated assemblies can strand product-specific capacity on either side. Metal additive manufacturing is more ambiguous than usually portrayed — direct-metal printing competes for some low-volume, high-complexity parts and is least economic against ordinary large steel castings, while printed sand molds and cores eliminate conventional tooling and can strengthen the foundries that adopt them, with the final part still a casting.[3][4]
• Quality escapes: defects on safety-critical parts trigger scrap, warranty, stoppages, recalls, penalties, or permanent disqualification; a single defect or furnace interruption can block an OEM's far larger assembly.[3][4]
• Environmental and safety liabilities — and they are quantified. Old plants carry air-control, waste, remediation, silica, molten-metal, and (nonferrous) dust-explosion exposure, and every grandchild with a published injury rate runs well above the private-industry benchmark of 2.3 recordable cases per 100 full-time workers: iron foundries 5.5 (2024), steel foundries except investment 5.6 (2023), die-casting 4.6 (2022), other nonferrous 4.5 (2022), and steel investment foundries 4.0 (2024). The years are mismatched and the figures are not strictly comparable, but the range is unambiguous — every one of the five is at least 1.7× the private-industry rate, and the ordering is intuitive: sand molding and heavy handling are more hazardous than lost-wax work.[3][4]
• Labor: experienced metallurgists, molders, welders, patternmakers, mold technicians, inspectors, and maintenance techs are hard to replace amid an aging skilled-trades base, and qualification work cannot simply be shifted to inexperienced labor.[3][4]
• Trade volatility: the same tariffs that protect domestic casting pricing raise steel, alloy, refractory, and equipment costs, and importers have historically absorbed most of the duty.[3][15]
• Classification ambiguity: a business marketed as an "aluminum foundry" may earn much of its revenue from die-casting, smelting, or finished-product manufacturing outside the intended code, and captive capacity is invisible in the federal counts.[3][4]
• Private-company leverage: sponsor-owned businesses add acquisition-debt, refinancing, and exit-timing risk on top of operating cyclicality.[3][4]

10. How to invest & outlook

Public routes are all indirect — there is no pure-play NAICS 3315 listing, even after the June 2026 Doncasters IPO widened access. Choose by segment exposure, not by whichever company mentions "casting," and never apply a foundry multiple to a diversified parent's consolidated earnings.

Note: die-casting (the largest nonferrous child by headcount) has no clean U.S. listing — exposure runs through the foreign listings above, plus a small captive inside Allison Transmission (ALSN), and should be valued on the parent's broader business.[4]

Private routes are where most of the industry sits: control or minority stakes, private credit, equipment finance, and sale-leasebacks in the family-, employee-, and PE-owned foundries named in Section 4. Diligence should center on plant-level evidence — saleable tons or good castings rather than melt capacity; utilization by furnace/press size; normalized yield and scrap/rework; contribution per mold or machine hour; pattern/die/mold ownership and condition; metal pass-through and escalator mechanics and working-capital lags; certification status (AS9100/Nadcap for aerospace); customer, program and end-market concentration; qualification-transfer risk and remaining program life; maintenance versus growth capital and the cost of melt modernization; environmental reserves and legacy sites; labor agreements; and debt headroom. Discount backlog that lacks firm quantities.[3][4]

Outlook (forward-looking judgment, not a reported forecast). The group is selective, not uniformly bullish, and the two halves point in different directions:

• The precision corner (spanning both children) has the strongest tailwind — record aircraft backlogs, the narrowbody ramp, defense budgets, data-center-driven gas-turbine demand, and Buy-American/DFARS content rules; Howmet guided to roughly $9.1 billion of 2026 revenue, about 10% growth, and the nonferrous superalloy corner is repricing hard (331529 output prices up ~11% in 2025 and 9% more by May 2026). But the cleanest large-cap proxies already price in much of that optimism, and the corner's own headwinds — skilled labor, energy and alloy volatility, aerospace air-pocket risk — are unresolved.[3][4]
• Ferrous is a two-track story: municipal-water replacement, domestic-content rules, defense, and mining wear support well-positioned iron and heavy-steel plants — Ampco-Pittsburgh reported first-half 2026 orders of about $268 million, up 32% from roughly $204 million — while EV adoption and soft ag/large-equipment demand pressure others, and iron's own output prices rose only ~2.6% year over year. Waupaca's simultaneous closure of a finishing operation and $100 million-plus modernization captures the divide.[3]
• Nonferrous is the most contested: aerospace/defense aluminum and superalloy are attractive; die-casting's automotive cyclicality, uneven EV adoption, heavy capital needs, and structurally thin reported margins (mid single digits at the listed comparables) will widen the gap between flexible, well-utilized plants and legacy ICE-dependent ones; undifferentiated conventional aluminum blocks and heads deserve a discount.[4]

Net: the best opportunities are plant-specific and child-specific — a qualified precision caster riding aerospace, a heavy-steel foundry with recurring wear-part demand, an efficient iron plant anchored to water infrastructure, or a flexible die-caster winning structural gigacasting work — rather than a uniform "foundry recovery." Three cautions this level exists to deliver. The group's low aggregate HHI should never be mistaken for a level playing field; the money and the moats are unevenly distributed, and the federal data hides that unevenness rather than revealing it — only one of six grandchildren publishes either a current receipts line or a concentration ratio. The reported margin ladder, from under 4% in iron conversion work to 33% in aerospace engine castings, is the sharpest evidence of where the value sits. And the near-term casting upcycle — half of it priced, not shipped — sits inside a projected 14.7% decline in this group's employment by 2034, so durable returns are more likely to come from utilization, casting yield, pricing discipline, and capital execution in a shrinking, qualification-protected market than from riding volume growth.[3][4][9][16]

Sources

1. U.S. Census Bureau, "2022 NAICS Definitions: Foundries (3315) and children (33151 Ferrous Metal Foundries; 33152 Nonferrous Metal Foundries)," 2022. https://www.census.gov/naics/
2. Histometrics ground-truth federal statistics, NAICS 3315: U.S. Census Bureau County Business Patterns (2023) — establishments (1,335), employment (107,273), annual payroll ($7.22B), first-quarter payroll ($1.84B); U.S. Census Bureau 2022 Economic Census concentration file — receipts ($32.76B), firms (1,211), CR4 20.2%, CR8 30.4%, CR20 48.2%, CR50 63.7%, HHI 170.4. https://www.census.gov/programs-surveys/cbp.html
3. Histometrics child primer, "U.S. Ferrous Metal Foundries (NAICS 33151)," synthesizing children 331511/331512/331513 and drawing on U.S. Census Bureau CBP (2023), the 2022 Economic Census (receipts $19.51B; firms 411; CR4 29.1%; HHI 337.4; steel-investment receipts $3.95B across 84 firms with top-four 73.5%), the 2017 Economic Census (iron shipments $8.584B), the 2002 Economic Census, and an OSHA final economic analysis (modeled 331513 firm count, revenue, and profit rate); Howmet Aerospace, Mueller Water Products, Caterpillar, Deere, Air T, Castings plc, Ampco-Pittsburgh, Weir Group, and Impro Precision filings; Doncasters (DPC Holdings) IPO and SEC filings; Berkshire Hathaway/Precision Castparts; Monomoy, Gamut, Warburg Pincus/Berkshire Partners, Charlotte Pipe, and TerraMar transaction releases; U.S. Energy Information Administration Manufacturing Energy Consumption Survey (2022); Federal Reserve Bank of St. Louis capacity-utilization and producer-price series; BLS producer-price, injury, and employment-projection data; American Foundry Society outlook survey; EPA, OSHA, SBA, DOJ, USITC, and Section 232 sources.
4. Histometrics child primer, "U.S. Nonferrous Metal Foundries (NAICS 33152)," synthesizing children 331523/331524/331529 and drawing on U.S. Census Bureau CBP (2023) and 2022 Economic Census (receipts $13.25B; firms 811; CR4 19.8%; HHI 177.6); Doncasters/DPC Holdings S-1 and June 2026 IPO; Howmet Aerospace, ATI, Brunswick, General Motors, and Allison Transmission filings; Berkshire Hathaway/Precision Castparts (including the Jenkintown fire disclosure); Ryobi, Nemak, Linamar, and Impro Precision filings; Warburg Pincus, Ares/Form Technologies, MiddleGround, Architect Equity, and Wisconsin Aluminum Foundry transaction releases; U.S. Energy Information Administration Manufacturing Energy Consumption Survey (2022); Bureau of Labor Statistics producer-price and injury data; North American Die Casting Association; U.S. Geological Survey Mineral Commodity Summaries 2026; U.S. Department of Energy lightweight-materials program; EPA, OSHA, SBA, and Section 232 sources.
5. Reconciliation of child figures to the group total: receipts $19.51B + $13.25B = $32.76B; employment 57,092 + 50,181 = 107,273; establishments 516 + 819 = 1,335 (all exact); annual payroll $3.97B + $3.243B ≈ $7.22B (to rounding); firms 411 + 811 = 1,222 vs. 1,211 group total (≈11 firms operate in both children). The steel-investment share of group receipts (~12%) is derived from $3.95B against the $32.76B group total. Derived from sources [2], [3], and [4].
6. U.S. Environmental Protection Agency, "Iron and Steel Foundries: NESHAP (40 CFR Part 63, Subparts EEEEE and ZZZZZ)," 2026. https://www.epa.gov/stationary-sources-air-pollution/iron-and-steel-foundries-national-emissions-standards-hazardous
7. U.S. Environmental Protection Agency, "Aluminum, Copper, and Other Nonferrous Foundries: NESHAP (40 CFR Part 63)," updated 2025. https://www.epa.gov/stationary-sources-air-pollution/aluminum-copper-and-other-nonferrous-foundries-national-emission
8. Occupational Safety and Health Administration, "Respirable Crystalline Silica in General Industry (29 CFR 1910.1053)," current. https://www.osha.gov/silica-crystalline/general-industry-maritime
9. U.S. Bureau of Labor Statistics, "Industry Employment and Output Projections" (NAICS 3315 foundries, 2024–2034: employment 106,000 to 90,400, −14.7%; real output $26.9B to $24.8B), current. https://www.bls.gov/emp/tables/industry-employment-and-output.htm
10. U.S. Small Business Administration, "Table of Small Business Size Standards," 2023. https://www.sba.gov/document/support-table-size-standards
11. U.S. Energy Information Administration, "2022 Manufacturing Energy Consumption Survey: Table 6.3 — Energy Intensity," 2024. https://www.eia.gov/consumption/manufacturing/data/2022/pdf/Table6_3.pdf
12. U.S. Environmental Protection Agency, "Beneficial Uses of Spent Foundry Sands," 2026. https://www.epa.gov/smm/beneficial-uses-spent-foundry-sands
13. U.S. Environmental Protection Agency, "Metal Molding and Casting (Foundries) Effluent Guidelines (40 CFR Part 464)," updated 2026. https://www.epa.gov/eg/metal-molding-and-casting-foundries-effluent-guidelines
14. The White House, "Further Adjusting the Tariff Regimes for Imports of Aluminum, Steel and Copper into the United States" (Section 232), 2026. https://www.whitehouse.gov/presidential-actions/2026/06/further-adjusting-the-tariff-regimes-for-imports-of-aluminum-steel-and-copper-into-the-united-states/
15. U.S. International Trade Commission, "Economic Impact of Section 232 and Section 301 Tariffs," current. https://www.usitc.gov/keywords/301_tariffs
16. American Foundry Society, "Quarterly Metalcasters Outlook Survey," March 2026. https://www.afsinc.org/news/2026/03/17/quarterly-metalcasters-outlook-survey-shows-increasing-optimism-among-foundry