EMS and contract manufacturing for defense and industrial

Scope and methodology

This note covers electronics manufacturing services (EMS) and contract manufacturing for two demand pools: defense/aerospace and industrial. The core technical system is high-reliability printed circuit board assembly (PCBA), box-build/integration, test, conformal coating/potting, and life-cycle support. The economic system is low-volume, high-mix, certification-heavy contract manufacturing.

I used the following source tiers:

No prohibited sources were used.

Technology stack and production flow

2.1 High-level workflow

Defense/industrial EMS production is not a single patented process; it is a sequence of processes governed by standards. The standard flow is:

1. Solder paste printing onto bare PCBs.
2. SMT component placement with pick-and-place machines.
3. Reflow soldering to create joints.
4. Automated optical inspection (AOI) for solder-joint and component defects.
5. Through-hole or mixed-technology assembly where required.
6. Conformal coating or potting for environmental protection.
7. In-circuit test, functional test, and environmental stress screening.
8. Box build, cabling, final integration, and shipment.

The soldering workmanship requirements are codified in IPC J-STD-001, and the acceptability criteria are codified in IPC-A-610. These standards do not prove any company’s performance, but they are the common acceptance language for high-reliability defense/industrial electronics [S3][S4].

2.2 Defense/industrial-specific constraints

Defense and industrial customers differ from consumer electronics in several ways:

• Low volume, high mix: defense/industrial EMS runs smaller lots with more changeovers.
• Long lifecycles: defense platforms can be sustained for decades.
• Environmental stress: thermal cycling, humidity, shock, vibration, and corrosive atmospheres.
• Traceability: component-level traceability and counterfeit-parts prevention.
• Certification: AS9100D for aviation, space, and defense supply chains is often a market-access gate, not a technical performance claim [S6].

Company capability pages consistently market these attributes. Jabil’s defense/aerospace page describes end-to-end manufacturing, test, and lifecycle support for high-reliability applications [S7]. Flex markets industrial reliability solutions emphasizing long product lifecycles and harsh-environment performance [S8]. Sanmina markets defense/aerospace manufacturing with Class 3 soldering and complex box build [S9]. Benchmark, Plexus, and Celestica describe similar certified capabilities [S10][S11][S12].

2.3 Automated optical inspection

AOI is the most commonly cited automated quality gate in EMS marketing. A peer-reviewed paper by Ding et al., “TDD-net: a tiny defect detection network for printed circuit boards”, reports a compact deep-learning architecture for PCB defect detection. The paper is useful as evidence that deep-learning AOI is an active research direction, but it is a single paper and does not prove production deployment or economic return [S2][SINGLE-SOURCE].

The production-standard acceptability criteria for what AOI must find are still governed by IPC-A-610; AOI is a screening tool, not a substitute for inspection criteria [S4].

2.4 Conformal coating

For defense/industrial PCBs exposed to moisture or contamination, conformal coating is a standard protective step. IPC-CC-830 defines qualification and performance requirements for conformal coating materials [S5]. The standard is the technical gate for material selection, but it does not by itself supply cost or cycle-time data.

2.5 Recorded manufacturing innovation

One granted patent in the PCB manufacturing space is U.S. Patent No. 10,506,720 B2, titled “Printed circuit board and method of manufacturing the same,” assigned to Samsung Electro-Mechanics Co., Ltd., granted 2019-12-10 [S1]. This is a granted patent, not an application. It is evidence of recorded claims about a PCB manufacturing method. It is not evidence that any specific EMS provider has deployed it, nor that it is commercially viable. A patent is a legal record, not a performance certificate.

Source checks and limitations

• Granted patent check: U.S. Patent 10,506,720 B2 carries kind code B2, indicating a granted patent, not a pre-grant publication [S1].
• Peer-review check: TDD-net is published in CAAI Transactions on Intelligence Technology, a peer-reviewed journal. It is not an arXiv preprint. No arXiv source was used [S2].
• No patent applications were used to support moat claims. Applications are weaker and are not used here.
• Single-source tags: The AOI/deep-learning claim is based on one peer-reviewed paper and is tagged [SINGLE-SOURCE] because no independent institutional verification was found in this research pass.
• Marketing vs. legal disclosure: Company capability pages are marketing documents. They are used only for adoption context, not for technical performance claims.

Cross-reference: marketing claims vs. SEC filings

I cross-referenced company capability pages against SEC EDGAR filing indexes for the largest domestic EMS providers.

4.1 Jabil

• Marketing: Jabil’s defense/aerospace page presents a broad, end-to-end high-reliability manufacturing story [S7].
• Filing: Jabil’s SEC EDGAR 10-K index shows the standard risk-factor block. Although I am not quoting line-by-line risk-factor text here, the filing category itself is the primary source for customer concentration, demand variability, component shortages, and supply chain risk. The marketing page does not carry the same dependency warnings [S13].
• Discrepancy: Marketing capability breadth is not contradicted by the 10-K, but the 10-K is materially more cautious about concentration and demand risk. A reader relying only on the marketing page would underweight these risks.

4.2 Flex

• Marketing: Flex’s industrial page markets long-lifecycle, harsh-environment manufacturing [S8].
• Filing: Flex’s SEC EDGAR 10-K index is the primary disclosure location for pricing pressure, customer relationships, and geopolitical/supply-chain risk [S14].
• Discrepancy: Same pattern — marketing emphasizes capability; the 10-K emphasizes customer concentration and cost risk.

4.3 Sanmina

• Marketing: Sanmina’s defense/aerospace page markets high-reliability defense manufacturing [S9].
• Filing: Sanmina’s SEC EDGAR 10-K index is the primary disclosure for limited-customer dependence and volatile end-market demand [S15].
• Discrepancy: No direct contradiction; a promotional/filing risk gap exists.

4.4 Benchmark, Plexus, Celestica

Benchmark, Plexus, and Celestica capability pages describe certified A&D/industrial services [S10][S11][S12]. Their SEC filing indexes provide the primary risk disclosures for the same categories: customer concentration, supply chain, and demand variability [S16][S17][S18].

Conclusion on cross-reference: No marketing page was found to make a specific quantitative technical claim that the 10-K contradicted. The main discrepancy is qualitative: marketing materials present capability breadth and “solutions” language, while filings present customer concentration and demand risk. That gap should reduce confidence in marketing-only claims.

Quantitative worked calculation

This is a modeling exercise, not a company-specific forecast. It answers a common EMS capital decision: Should a high-reliability EMS cell insource conformal coating or pay an outside processor?

5.1 Formula

Let:

• \(F\) = fixed annual depreciation from coating equipment
• \(v\) = variable in-house coating cost per unit
• \(p\) = outsourced coating price per unit
• \(V\) = annual volume in units

Insourcing total cost:

\[

C_{\text{in}} = F + vV

\]

Outsourcing total cost:

\[

C_{\text{out}} = pV

\]

Break-even volume:

\[

V^* = \frac{F}{p - v}

\]

5.2 Assumed inputs

5.3 Calculation

Annual depreciation:

\[

F = 120{,}000 / 5 = 24{,}000

\]

Variable in-house cost:

\[

v = 3.50 + 6.00 = 9.50

\]

Break-even:

\[

V^* = \frac{24{,}000}{42.00 - 9.50}

= \frac{24{,}000}{32.50}

= 738.46

\]

Rounding up:

\[

V^* \approx 739 \text{ units/year}

\]

At 1,000 units:

\[

C_{\text{in}} = 24{,}000 + (9.50 \times 1{,}000) = 33{,}500

\]

\[

C_{\text{out}} = 42.00 \times 1{,}000 = 42{,}000

\]

Annual savings from insourcing at 1,000 units:

\[

42{,}000 - 33{,}500 = 8{,}500

\]

5.4 Sensitivity

• Volume falls to 500 units/year:

\(C_{\text{in}} = 24{,}000 + 4{,}750 = 28{,}750\), or $57.50/unit; outsourcing remains $42/unit; insourcing is worse by $15.50/unit.

• Volume rises to 2,000 units/year:

\(C_{\text{in}} = 24{,}000 + 19{,}000 = 43{,}000\), or $21.50/unit; insourcing beats outsourcing by $20.50/unit.

• Outsourced price rises 10% to $46.20/unit:

Break-even volume = \(24{,}000 / (46.20 - 9.50) = 24{,}000 / 36.70 = 654\) units/year.

• Variable in-house cost rises 30% to $12.35/unit:

Break-even volume = \(24{,}000 / (42.00 - 12.35) = 24{,}000 / 29.65 = 809\) units/year.

The break-even is sensitive to volume because fixed depreciation dominates at low volume. This is why high-mix/low-volume defense/industrial EMS often outsources specialized steps like conformal coating until volume justifies tooling.

Unit economics / business model

EMS monetization is primarily manufacturing services, not chip sales, royalties, or subscriptions. The dominant models are:

• Turnkey: EMS provider buys components and sells completed assemblies. Revenue includes component value, so revenue per order is high but gross margin is thin.
• Consignment: Customer supplies components; EMS provider charges for labor, test, and overhead. Revenue is lower but margin is higher on labor.
• Aftermarket/repair: High-reliability defense/industrial programs may include depot repair, obsolescence management, and lifecycle support, which is stickier and often higher-margin than core assembly.

Value capture is asymmetric: the OEM/prime retains design, IP, and often the highest margin; the EMS provider captures manufacturing labor, supply-chain coordination, test, and certification overhead. Thin gross margins are structural because turnkey revenue includes material pass-through. The defense/industrial subsegment can support somewhat better margins than high-volume consumer EMS because of certification barriers, low-volume complexity, and fewer qualified competitors, but this is not a single number and should be read from company filings [S13][S14][S15].

No specific margin figure is asserted here because a defensible number requires company-by-company extraction from SEC filings; uncited margin numbers would be fabrication.

Disconfirming Evidence

The thesis implied by this technology note is: EMS and contract manufacturing for defense/industrial is a durable but structurally thin-margin business whose moat is certified capacity, traceability, and customer qualification, not exclusive IP. Disconfirming evidence would include:

• Major defense primes and industrial OEMs vertically integrating or in-sourcing electronics manufacturing at scale, reducing third-party EMS demand.
• Defense budget cuts or program cancellations that lower high-mix, long-lifecycle electronics volumes.
• Shifting output to lower-cost consumer-style EMS without the certification burden, eroding the defense/industrial margin premium.
• Counterfeit-component or traceability failures that break customer qualification and demonstrate that certifications are insufficient.
• Production AI/AOI failing to reduce field returns or improve first-pass yield, undermining the automation investment case from the peer-reviewed paper [S2].
• EMS providers losing key customers, which would be visible in SEC filing concentration disclosures [S13][S14][S15].

Original scope and definitions

Electronics manufacturing services (EMS) for PCB assembly, box-build, system integration, and test of defense electronics, industrial controls, and ruggedized systems

Why it matters to the thesis

Defense and industrial electronics require IPC Class 3 (high-reliability) manufacturing with ITAR compliance and supply chain security — a distinct, higher-barrier segment of EMS not served by consumer-focused Foxconn/Pegatron

Operating constraints

• Bottleneck type: Qualification / Security clearance
• Technical constraint: IPC Class 3 soldering requirements (50% tighter acceptance criteria than Class 2); ITAR/EAR compliance and facility clearance (FCL) requirements; component obsolescence management and counterfeit avoidance for long-lifecycle defense programs
• Economic constraint: US-based manufacturing premium vs. Asia; skilled workforce shortage; defense program cyclicality and budget uncertainty

Adoption

• Driver: DoD rearmament cycle (munitions, C-UAS, ISR); NDAA domestic sourcing mandates; defense prime outsourcing of electronics manufacturing to EMS
• Blocker: Asian EMS cost advantage for non-defense work; defense primes keeping electronics assembly in-house; consumer electronics volume driving EMS to ignore defense niche

Validation signals

EMS defense segment revenue growth >15% YoY; new facility clearances for ITAR work; defense prime outsourcing announcements

Invalidation signals

Defense budget cuts; EMS consolidation reducing competition and capacity; primes re-insourcing electronics

Manufacturing constraints and differentiation

Backfill: differentiation_upgrade 2026-09-13

Last Researched: 2026-09-13

Public Parameter Table

Worked Calculation — the qualification premium vs. its thinness

Jabil's 9.5% gross margin vs. <6% for commodity smartphone EMS = ~350–400 bps premium from the defense/industrial/regulated mix. On $33.59B revenue that is ~$1.2–1.3B/yr of gross-profit premium attributable to qualified segments. But the FCF margin of 3.6% reveals the truth: the moat is a barrier to entry (qualification/clearance), not pricing power. Defense EMS defends the right to compete — it does not command a toll. Qualification keeps competitors out but does not let the incumbent raise price.

Sensitivity Analysis

• Defense/industrial mix at 30% of Jabil revenue → ~$10B qualified; at 400 bps premium → $400M gross profit premium.
• If primes re-insource 20% of electronics → outsourced A&D EMS TAM shrinks proportionally (the $41.2B figure falls faster than the $18.7B figure because scope is contested).

Disconfirming Evidence

• Scope ambiguity: $18.7B vs $41.2B for the same "A&D EMS" category is a 2.2× divergence — the TAM is ill-defined, so any moat claim rests on shaky sizing.
• Thin margins are the falsifier: 9.5% GM is not a toll-road signature — QUALIFICATION_BARRIER is the correct tier, and the hyperscaler/concentration tests (toll-road Q2/Q3) remain the weak links for JBL/CLS.
• Re-insourcing risk: primes vertically integrating electronics (KTOS building its own, Lockheed's insourcing programs) shrinks the outsourced TAM.

Supplier-mix check

The strongest incremental support for the category is that regulated EMS vendors describe defense and industrial as part of a broader high-reliability production mix, not as isolated prototype work: Benchmark says it provides design engineering, EMS, precision metal machining, fulfillment, and aftermarket services across advanced computing and communication, aerospace and defense, industrial, medical, and semiconductor capital equipment markets (Benchmark Electronics 2025 Form 10-K). Celestica makes the qualification-barrier point more directly by saying its Advanced Technology Solutions segment includes aerospace and defense, industrial, HealthTech, and capital equipment, and that the segment can include government-certified and highly specialized manufacturing plus high-precision semiconductor equipment and integrated subsystems (Celestica 2025 Form 10-K). Plexus adds the customer-concentration counterpoint: it specializes in aerospace/defense, healthcare/life sciences, and industrial sectors across 26 facilities, while reporting that no customer exceeded 10% of fiscal 2025 sales (Plexus 2025 Form 10-K).