Printed Circuit Assembly (Electronic Assembly) Manufacturing — U.S. Industry Primer (NAICS 334418)

1. Overview

Almost every electronic product — a car's braking computer, an insulin pump, a fighter jet's radar, an AI server — runs on a printed circuit assembly (PCA): a bare circuit board with the chips, resistors, connectors and other parts soldered onto it. NAICS code 334418 covers the U.S. establishments whose main business is doing that soldering and assembly work: loading components onto boards and shipping the finished, populated boards [1]. (NAICS is the North American Industry Classification System, the government's standard scheme for grouping businesses.)

Most of these firms are contract manufacturers — they don't own the product design or the brand; they build boards (and often whole devices) to another company's specification. In investor shorthand the business is called EMS (electronics manufacturing services) or ECM (electronic contract manufacturing). That structure is the single most important thing to understand about the industry: owners make money not from a proprietary product with a fat markup, but from running factories at high utilization on razor-thin margins, at very large volume.

Why an investor cares: this is the physical backbone of the entire electronics economy, and demand tracks the biggest capital-spending waves in the market — AI data centers, electric vehicles, defense modernization, medical devices. It is cyclical, capital-intensive and low-margin, which makes it a classic operating-leverage play: small swings in factory volume produce large swings in profit.

• Public-market route: a clean set of listed contract manufacturers (Jabil, Flex, Celestica, Sanmina, Plexus, Benchmark, and smaller names), plus the Taiwan-listed giants that dominate the global business (Foxconn, Pegatron).
• Private route: several hundred privately held U.S. "board shops" and job shops, foreign-owned plants operating in the U.S., and the private-equity roll-ups that are steadily buying them up.

2. What it is and how it's structured

In scope (334418): establishments that load components onto printed circuit boards and manufacture/ship the loaded boards — also called printed circuit assemblies, electronic assemblies or modules [1]. The core factory process is SMT (surface-mount technology): solder paste is printed through a stencil onto the board, automated pick-and-place machines position surface-mount components, the board passes through a reflow-solder oven, through-hole components are inserted and wave- or hand-soldered, and the assembly is inspected (often by automated optical inspection or X-ray) and tested [2]. Higher-mix shops add conformal coating, box build (assembling the finished enclosure) and testing. IPC's principal assembly standards — J-STD-001 for soldering process control and IPC-A-610 for post-assembly acceptance — govern the workmanship [3].

Explicitly excluded — and the adjacent NAICS codes that catch it:

• Making the bare board (etched copper, no parts yet) → 334412, Bare Printed Circuit Board Manufacturing [1].
• Making the raw laminate the board is built from → 334419, Other Electronic Component Manufacturing [1].
• Making telephone system modules → 334210, Telephone Apparatus Manufacturing [1].
• Building a finished product that happens to contain a loaded board (a laptop, a router, an MRI machine) → classified wherever that final product lives, based on the end-product process [1].

That last exclusion matters for reading the size figures: when a contract manufacturer builds a complete device rather than shipping loose assembled boards, the government may classify that plant under the end-product code, not 334418.

Ownership mix. Domestically the industry is a barbell. A handful of large plants (many owned by the big multinational EMS firms or by foreign parents) sit at one end; several hundred small, privately owned job shops sit at the other. Federal data show 820 firms operating 861 establishments in the U.S. [4][5] — meaning the typical firm runs a single plant. Average plant size is roughly 58 employees, and average revenue per firm is about $30 million (both figures derived from the Census counts below) [4][5] — i.e. mostly small and mid-sized businesses, with a few very large operations on top. Nearly all are for-profit corporations; there is essentially no government or nonprofit ownership.

3. How big it is

U.S. federal statistics for NAICS 334418:

Concentration is moderate for a manufacturing industry: the top 4 firms account for 43.6% of U.S. revenue, the top 8 for 52%, the top 20 for 62.6%, and the top 50 for 72.9% [4]. (The Herfindahl-Hirschman Index, the standard antitrust concentration measure, is suppressed in the Census release, so we do not state a value for it [4].)

The undercount caveat — read this before comparing to the companies below. The domestic revenue figure counts only assembly performed on U.S. soil. It dramatically understates the economic weight of U.S.-based EMS companies, because the giants do most of their manufacturing offshore. Jabil reported 75% of its fiscal-2025 revenue as foreign-source [8]; Flex's fiscal-2025 geographic breakdown shows $6.9 billion produced in Mexico and $4.3 billion in China [9]. So 334418 measures the U.S. factory footprint of board assembly (which skews toward regulated, U.S.-only work like defense and medical), not the size of the American EMS industry as a corporate sector. Both readings are valid; just don't confuse them.

Data note: Beginning with its 2025 benchmark changes, BLS combined bare-board manufacturing and printed-circuit assembly into one published employment series covering NAICS 334412 and 334418 together; the combined figure cannot be presented as employment in 334418 alone [10].

4. The investable universe

Unusually for a "components" niche, there is a deep public bench — though the two largest players are listed in Taiwan, not the U.S. Revenue figures are each company's most recent full fiscal year and are global, not U.S.-only.

Major private and other owners. The domestic long tail is overwhelmingly private: several hundred U.S. board shops (examples include Zentech, SMTC, and dozens of regional job shops), foreign-owned plants operating on U.S. soil (Foxconn's Wisconsin site; Germany's Zollner; Singapore's Venture Corp and Venture-style Asian EMS firms), and a growing cohort of private-equity roll-ups consolidating small shops. For non-public investors the action is here — buying, building or backing regulated-market job shops — because the public names are dominated by the low-margin, high-volume end.

5. How the money works

This is a manufacturing business, so the economics are about volume, utilization and working capital, not pricing power. The levers owners actually pull:

• Gross margin is structurally thin. Because customers own the design and can re-bid the work, and because much reported revenue is the resale of customer-specific components, EMS gross margins typically run in the high single digits to low teens [19]. Fiscal-2025 gross margins were 8.9% at Jabil, 8.4% at Flex, 8.8% at Sanmina, 10.1% at Plexus and 7.0% at Kimball Electronics [8][9][13][14][18]. Corresponding operating margins were 4.0% at Jabil, 4.5% at Flex, 5.0% at Plexus and 3.1% at Kimball [8][9][14][18]. Double-digit operating margins are almost unheard of in this business [19].
• Capacity utilization is the profit switch. SMT lines cost millions and largely run whether busy or idle. Fill them and fixed cost spreads over more boards, so incremental volume drops disproportionately to profit; let them run half-empty in a downturn and margins collapse. Kimball's fiscal-2025 illustrates the leverage vividly: sales fell 13%, but gross profit fell 26% and gross margin declined from 8.2% to 7.0%, which the company attributed partly to lost absorption [18]. This operating leverage is why the stocks are so cyclical.
• Pass-through model on materials. Components are the majority of cost and are largely passed through to the customer, so revenue inflates and deflates with chip prices without necessarily changing dollar profit. The value-add owners actually capture is the labor, overhead and engineering on top — which is why gross profit dollars and margin, not headline revenue, are what to watch.
• Working capital and inventory turns. EMS firms front the cash for components before the customer pays, so inventory turns, days of inventory and free cash flow are core health metrics. Supply agreements may make customers responsible for authorized excess material, but recovery can be delayed or disputed. Sanmina notes its major supply agreements generally run three to five years but do not obligate customers to purchase minimum quantities [13]. The 2021-22 shortage era forced firms to stockpile parts; 2023 was a painful destocking year as customers worked inventory back down [20], directly cutting EMS revenue.
• Book-to-bill and backlog. The near-term demand gauge is the book-to-bill ratio (new orders received divided by product shipped/billed); above 1.0 signals a growing backlog. North American EMS book-to-bill ran above 1.0 in early 2024, an expansion signal [21].
• Customer concentration is the hidden risk in the margin. Big EMS firms often derive a large share of revenue from a few anchor customers. Jabil's five largest customers produced 36% of fiscal-2025 revenue, with one customer representing 16% [8]. Sanmina's ten largest customers represented 52% of fiscal-2025 sales [13]. Kimball's two disclosed large customers represented 19% and 11% of fiscal-2025 sales [18]. That wins volume but hands the customer pricing leverage and makes a single lost program materially painful.

The strategic escape from thin margins is going up-market: high-mix, low-volume work in regulated end-markets (aerospace, defense, medical, semiconductor capital equipment) that demands certifications, traceability and reliability, carries better margins, and is much harder to offshore. That is exactly where the U.S. plants and the mid-cap public names (Benchmark, Plexus, Sanmina) have positioned.

6. What drives demand

EMS is a derived-demand business: it grows when its customers' end-markets spend. The forward-looking growth drivers the industry is betting on:

• AI and data-center build-out. The largest near-term swing factor. Hyperscale cloud operators' capital spending on AI servers, networking switches and optical interconnects is pulling in specialist assemblers; Jabil's fiscal-2025 Intelligent Infrastructure revenue rose 34%, including a 30% increase from existing cloud and data-center customers [8]. Celestica's data-center segment drove ~21% revenue growth in 2024 and a sharply raised 2026 outlook [12], and Fabrinet rides the same AI-optics wave [15]. Flex reported cloud strength offsetting weaker non-cloud demand [9].
• Automotive electrification. Electric vehicles, advanced driver-assistance systems and connected-car platforms sharply raise the electronic content per vehicle, a durable multi-year tailwind [20].
• Medical devices — wearables, diagnostics, implantables — a high-margin, high-regulation niche EMS firms are actively expanding into [20].
• Industrial automation, IoT and 5G — sensors, controllers and connected industrial gear [20].
• Aerospace and defense. Rising defense budgets and munitions/electronics restocking; must be built in ITAR-registered U.S. plants (see below), which favors domestic capacity [22]. In 2024 the Department of Defense awarded $11.7 million under the Defense Production Act to expand U.S. PCBA capacity for hypersonic systems — evidence of strategic demand, not a broad subsidy, but a signal of where policy is pointing [23].
• Reshoring and nearshoring. Tariffs and supply-chain shocks are pushing production out of China. Full reshoring to the U.S. remains the minority choice; nearshoring to Mexico under the USMCA (United States-Mexico-Canada Agreement) is the larger flow — Flex's geographic mix, with $6.9 billion produced in Mexico versus $4.3 billion in China, illustrates this reality [9]. U.S. capacity is more likely to win work where intellectual-property protection, supply-chain resilience, engineering proximity, export controls or trusted-source requirements outweigh the domestic cost premium. U.S. incentives such as the CHIPS Act add a domestic pull [19][24].

Analysts size the global EMS market at roughly $648 billion in 2025, projected toward $850 billion by 2030 [20] — a mid-single-digit growth outlook, faster in the AI, automotive and medical segments.

7. Regulation

There is no single sector regulator; instead a stack of quality, export and environmental rules — and a plant's certifications largely determine which lucrative work it can win.

• ITAR (International Traffic in Arms Regulations). U.S. rules governing defense-related articles on the U.S. Munitions List; any board built for military or defense applications must be assembled in an ITAR-registered U.S. facility by U.S. persons [22]. This is the single biggest reason meaningful defense assembly stays onshore.
• EAR (Export Administration Regulations). BIS rules governing dual-use items; defense and certain high-technology programs may bring EAR controls and trusted-supplier requirements [25].
• Quality-management certifications. AS9100 (aerospace quality standard, built on ISO 9001) and ISO 13485 (medical devices) are effectively table stakes to bid regulated work [22]. The FDA's Quality Management System Regulation (QMSR), which incorporates ISO 13485:2016, became effective February 2, 2026 [26]; medical-device contract manufacturing can fall within FDA inspections and quality-system obligations.
• IPC workmanship standards. IPC-A-610 defines acceptability classes (Class 1 general, Class 2 dedicated-service, Class 3 high-reliability for aerospace/defense/life-critical), and J-STD-001 governs soldering process and materials [3][22]. MIL-PRF-31032 and Nadcap apply to defense-grade work [22].
• Cybersecurity/data. NIST SP 800-171 (protecting controlled unclassified information) is increasingly required on defense contracts [22].
• Environmental / materials. RoHS (EU Restriction of Hazardous Substances, limiting lead and other substances in electronics), REACH (EU chemicals rules) and U.S. conflict-minerals disclosure shape which materials and solders can be used and impose supply-chain reporting. EPA includes NAICS 334418 as a potentially covered industry under electrical and electronic component effluent guidelines, though actual coverage depends on the facility's processes [27].

8. Competitive dynamics and consolidation

Globally the industry is top-heavy: Foxconn alone is larger than every other player combined for high-volume consumer work, and the top handful control over half the world market [11]. Domestically, as the Census figures show, it is more fragmented — a top-4 U.S. share of 43.6% and a long private tail [4].

The defining structural split is low-mix/high-volume versus high-mix/high-reliability. Commodity, high-volume consumer assembly has migrated to Asia, where scale and labor cost win; that work is brutally price-competitive and low-margin. What remains and grows in the U.S. is high-mix, regulated, lower-volume work where certification, engineering support, speed and traceability matter more than headline labor cost [19].

Substitution comes from three directions: OEMs can manufacture internally; global EMS firms can take regional work; and original-design manufacturers (ODMs) can combine design and manufacturing at lower cost. Benchmark notes ODMs have gained share particularly in commoditized computing and telecommunications [16]. Domestic specialists are best protected where engineering interaction, regulatory qualification, traceability, security or low-volume complexity matters.

Consolidation runs on two tracks. Among the public majors, M&A is used to buy capabilities and end-market exposure (optical, RF, medical, semi-cap). Among the private long tail, private-equity roll-ups are steadily acquiring small board shops to build regional platforms with scale — a live thesis for private investors. Meanwhile the largest firms are actively reshaping their footprints toward Mexico and select U.S. sites to serve nearshoring demand and de-risk from China [19][24].

9. Risks

• Cyclicality and inventory whipsaw. Demand amplifies the swings of every end-market it serves; the 2023 destocking cut revenue across the group [20]. Operating leverage cuts both ways — Kimball's FY25 shows how a 13% revenue decline can translate to a 26% gross-profit decline [18].
• Thin margins, little pricing power. Because customers own the designs, work can be re-bid; a few points of gross margin is the whole game, and input-cost or wage inflation eats it quickly [19]. Flex warns that tariff pass-through can raise sales while lowering the reported operating-margin percentage and delaying operating cash flow [9].
• Customer concentration. Losing a single anchor program can dent a plant's economics materially. The disclosed concentration levels at Jabil (16% from one customer), Sanmina (52% from top 10) and Kimball (30% from top 2) illustrate the exposure [8][13][18].
• Component supply shocks. A shortage of one chip can idle an otherwise-full line; the industry has limited buffer and, in ingestion terms, "no retry" — it simply cannot ship the board. Shortages increase purchase premiums and buffer inventories; a demand correction then produces excess and obsolete inventory, weaker cash conversion and write-offs.
• AI-demand concentration. The current growth story leans heavily on hyperscaler AI capex. If that spending normalizes, the names most levered to it (Celestica, Fabrinet, parts of Jabil/Flex) are most exposed — a forward-looking, not settled, risk.
• Trade and tariff policy. The same tariffs creating nearshoring opportunity also raise input costs and inject planning uncertainty [19].
• Geographic/geopolitical. Heavy reliance on Asian and Mexican plants exposes the majors to China-U.S. tension and cross-border disruption.
• Capital intensity. Keeping SMT lines current requires continuous reinvestment; falling behind on capability loses the high-margin work.
• Labor. The risk is less about generic assembly-headcount availability than about retaining process engineers, quality personnel, test technicians and certified soldering or rework specialists. Loss of experienced operators can lower yield and delivery performance.

10. How to invest and the outlook

Public-market routes. The listed pure-plays give direct exposure across the risk spectrum:

• AI/data-center leverage: Celestica (CLS) and Fabrinet (FN) are the most direct plays on hyperscaler AI hardware and optics [12][15] — highest growth, highest sensitivity if AI capex cools.
• Diversified scale: Jabil (JBL) and Flex (FLEX), the two U.S.-run giants, offer broad end-market exposure at large scale [8][9].
• Regulated, higher-margin mix: Sanmina (SANM), Plexus (PLXS) and Benchmark (BHE) concentrate on aerospace/defense, medical and semiconductor-capital-equipment work with better margins and stickier customers [13][14][16].
• Global volume: Foxconn (2317.TW) and Pegatron (4938.TW) for those able to buy Taiwan listings [11]. There is no dedicated U.S.-listed EMS ETF; exposure otherwise comes bundled inside broad semiconductor or industrial-technology funds. Because the business is cyclical and low-margin, valuation is usually framed on earnings/cash-flow multiples and the point in the demand cycle rather than on yield — dividends across the group are modest to nil, with capital more often returned via buybacks.

Private routes. For non-public investors the opportunity is concentrated in the fragmented domestic tail: acquiring, building or backing regulated-market job shops (defense/ITAR, medical, semi-cap) where certifications create a moat and margins beat the commodity end, and participating in the private-equity roll-up of small board shops into regional platforms. Foreign strategics buying U.S. capacity for nearshoring and ITAR access are another live channel. The fragmented establishment base — 861 plants, mostly single-site firms — makes add-on consolidation plausible, particularly when a buyer can centralize purchasing, sales and back-office functions while preserving customer-specific certifications and engineering relationships.

Private-acquisition diligence. Critical underwriting items include revenue and gross profit by customer and program; program life and renewal history; excess-material recovery provisions; customer-owned versus company-owned inventory; backlog conversion; equipment age and utilization; first-pass yield, scrap and rework; certifications and audit history; dependence on individual engineers; environmental liabilities; and the capital required to support awarded programs. Headline EBITDA is unsafe if it was produced by temporarily elevated component premiums, customer-funded inventory, underinvestment in equipment or a program nearing end of life.

Near-term outlook (forward-looking). The demand set-up is favorable: AI-hardware build-out, rising electronic content in vehicles, defense restocking, and a structural reshoring/nearshoring shift toward Mexico and select U.S. sites all point the same direction, with book-to-bill signaling expansion entering the period [12][20][21][24]. The offsetting judgment calls are how much of the current strength is AI-capex concentration that could normalize, whether tariff policy nets out as tailwind or cost, and the industry's permanent constraint — thin margins that leave little cushion when the cycle turns. Net: a structurally growing, strategically important industry, but a cyclical, low-margin one where execution on utilization and mix, not top-line growth alone, separates the winners.

---

Sources

1. U.S. Census Bureau. "2022 NAICS Definition — 334418 Printed Circuit Assembly (Electronic Assembly) Manufacturing." 2022. https://www.census.gov/naics/?input=334418&year=2022
2. U.S. EPA. "Printed Wiring Board Industry — Process Description." https://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=30003PO0.TXT
3. IPC. "IPC Releases Revisions to Two Leading Standards for Electronics Assembly (J-STD-001 and IPC-A-610)." https://www.ipc.org/news-release/ipc-releases-j-revisions-two-leading-standards-electronics-assembly
4. U.S. Census Bureau. "2022 Economic Census — Concentration and Receipts, NAICS 334418." 2022. https://www.census.gov/programs-surveys/economic-census.html
5. U.S. Census Bureau. "County Business Patterns (CBP), 2023 — NAICS 334418." 2023. https://www.census.gov/programs-surveys/cbp.html
6. U.S. Census Bureau. "Annual Integrated Economic Survey — NAICS 334418, 2023." https://data.census.gov/table/AIESEXP01TIMESERIES.AIES00EXP01?codeset=naics~334418&g=010XX00US
7. U.S. Small Business Administration. "Table of Small Business Size Standards (NAICS 334418 = 750 employees)." 2023. https://www.sba.gov/document/support-table-size-standards
8. Jabil Inc. "Fiscal Year 2025 Form 10-K." 2025. https://www.sec.gov/Archives/edgar/data/898293/000162828025045293/jbl-20250831.htm
9. Flex Ltd. "Fiscal Year 2025 Form 10-K." 2025. https://www.sec.gov/Archives/edgar/data/866374/000086637425000027/flex-20250331.htm
10. U.S. Bureau of Labor Statistics. "2025 Benchmark Series Changes — NAICS 334412/334418." 2025. https://www.bls.gov/ces/notices/2025/2025-bmk-series-changes.htm
11. Karkhana.io. "Top 10 Largest Electronics Manufacturing Services (EMS) Companies in the World in 2025." 2025. https://karkhana.io/exploring-the-top-10-largest-electronics-manufacturing-services-ems-companies-in-the-world-in-2025/
12. Kavout. "Is Celestica Poised to Dominate the AI Infrastructure Boom" (Celestica FY2024 revenue ~$9.6B, +21%; 2026 outlook). 2025. https://www.kavout.com/market-lens/is-celestica-poised-to-dominate-the-ai-infrastructure-boom
13. Sanmina Corp. "Fiscal 2025 Form 10-K." 2025. https://www.sec.gov/Archives/edgar/data/897723/000089772325000042/sanm-20250927.htm
14. Plexus Corp. "Fiscal Year 2025 Form 10-K." 2025. https://www.sec.gov/Archives/edgar/data/785786/000078578625000064/plxs-20250927.htm
15. Fabrinet. "Fourth Quarter and Fiscal Year 2024 Financial Results." 2024. https://investor.fabrinet.com/news-releases/news-release-details/fabrinet-announces-fourth-quarter-and-fiscal-year-2024-financial
16. Benchmark Electronics. "Fiscal Year 2025 Form 10-K." 2025. https://www.sec.gov/Archives/edgar/data/863436/000119312526064849/bhe-20251231.htm
17. TTM Technologies. "Fourth Quarter and Fiscal Year 2024 Results." 2025. https://www.globenewswire.com/news-release/2025/02/05/3021577/0/en/TTM-Technologies-Inc-Reports-Fourth-Quarter-and-Fiscal-Year-2024-Results.html
18. Kimball Electronics. "Fiscal Year 2025 Form 10-K." 2025. https://www.sec.gov/Archives/edgar/data/1606757/000160675725000027/ke-20250630.htm
19. Partstat. "EMS Providers and the Struggle Against Dwindling Profit Margins." 2023. https://blog.partstat.com/ems-providers-and-the-struggle-against-dwindling-profit-margins/
20. Fortune Business Insights. "Electronic Manufacturing Services (EMS) Market Size, Forecast [2034]." 2025. https://www.fortunebusinessinsights.com/electronic-manufacturing-services-ems-market-105519
21. Global SMT & Packaging. "EMS Book-to-Bill (North American EMS shipments and orders)." 2024. https://www.globalsmt.net/world-news/ems-book-to-bill-up-2-6-percent-in-january/
22. Midwest PCB (MPCS, Inc.). "Why PCB Certifications Matter: AS9100D, MIL-PRF-31032, ISO, UL, ITAR and IPC Standards." 2026. https://www.midwestpcb.com/why-pcb-certifications-matter-understanding-as9100d-mil-prf-31032-iso-ul-and-ipc-standards/
23. U.S. Department of Defense. "Department of Defense to Expand Manufacturing of Printed Circuit Board Assemblies." 2024. https://www.defense.gov/News/Releases/Release/Article/3701760/department-of-defense-to-expand-manufacturing-of-printed-circuit-board-assembli/
24. TeamSMT. "Tariffs, Reshoring, and Supply Chain Strategies in the EMS Industry." 2025. https://www.teamsmt.com/blog/2025-tariffs-reshoring-ems-industry
25. U.S. Bureau of Industry and Security. "Export Administration Regulations (EAR)." https://www.bis.gov/regulations/ear
26. U.S. Food and Drug Administration. "Quality Management System Regulation (QMSR)." 2026. https://www.fda.gov/medical-devices/postmarket-requirements-devices/quality-management-system-regulation-qmsr
27. U.S. EPA. "Electrical and Electronic Components Effluent Guidelines." https://www.epa.gov/eg/electrical-and-electronic-components-effluent-guidelines