ELTK
ELTKBusiness Overview
Step 01 — Business Model, Value Chain, and Unit Economics
Company: Eltek Ltd (ELTK) | Date: 2026-06-17 | Sector track: General Corporate (PCB manufacturer)
1. Key Findings
- Eltek is a qualification-gated, high-mix / low-volume custom PCB fabricator — not a commodity board maker. It builds complex rigid, multilayer (up to 40 layers), and especially flex-rigid and HDI boards in short runs, prototypes, and low-to-medium volumes for customers who pay for certification, trust, and engineering — not unit price [S1]. Revenue = boards shipped × price; there is no subscription/recurring layer.
- The business has effectively become an Israeli defense-PCB play. Defense & aerospace rose from 51% → 65% → 73% of production FY2023–FY2025 [S1]; Israel is now ~68% of revenue (up as domestic defense demand surged) [S1]. Medical (7%), industrial (9%), and distributors/CEMs (11%) are the rest [S1].
- Unit economics are dominated by four levers, in order: (1) capacity utilization / fixed-cost absorption, (2) product mix (flex-rigid & high-layer = high margin), (3) labor cost & yield, (4) FX (NIS cost base vs. USD reporting). Management quantified the operating leverage explicitly: at current scale, incremental revenue drops ~$0.50 on the dollar to gross profit [S3]. That cuts both ways — it is why margins swung from 28% (FY2023) to a gross loss (Q1 2026).
- Switching costs are real but program-level, not contractual. Each board is qualified to a customer program (ITAR since 2009, AS9100, Nadcap, US DoD QPL MIL-PRF-55110G/50884E) [S1]; requalifying a flight/defense board elsewhere is costly and slow, so repeat orders within a program are sticky — but there is no RPO, ARR, or take-or-pay contract. Backlog is simply POs scheduled within 24 months [S1].
- Capital intensity is the structural reality. A single ~90,000 sq ft Petach-Tikva plant (lease to 2039) [S1] running a limited set of large, often one-of-a-kind machines; ~$17M invested over three years to upgrade/expand [S1]. A machine failure can halt production for days-to-months — flagged as the #1 risk factor [S1].
Net for thesis: Mixed — a genuinely differentiated, defensible niche model (the source of the bull case) wrapped around a capital-intensive, single-plant, FX-exposed cost structure with brutal operating leverage (the source of the trough).
2. Implications for Thesis and Valuation
- The model is the reason the trough is plausibly transitory: in a high-mix fab, gross margin is mostly a function of how much qualified revenue runs across a largely fixed cost base. The FY2023 print (28.1% GM) shows what the same plant earns when utilization and mix are favorable; FY2025/Q1 2026 shows the same plant under-absorbed mid-relocation [S1][S2]. Valuation hinges on which state is "normal."
- It is also the reason the downside is real: there is no recurring revenue to cushion a demand air-pocket, the cost base is sticky (skilled labor, NIS-denominated), and FX is exogenous. A single-plant, single-geography concentration caps the multiple.
- Metrics that matter (feed Step 05 KPI selection): capacity-utilization proxy, gross margin, product/end-market mix (% flex-rigid, % defense), backlog (with the 24-month caveat), ASP, revenue, and labor cost per unit. Metrics that are irrelevant here: NRR/ARR/MAU/DAU (no subscription), TPV/take-rate (not payments), same-store sales (not retail), AUM (not finance). Forcing them would be analytical noise.
3. Objective
Explain how Eltek actually makes money — products, customers, pricing, sales motion, the value chain and its switching points, and the unit economics that drive the P&L — before layering on market structure (Step 02) and valuation.
4. Narrative Analysis
What it sells and to whom. Eltek manufactures custom-designed PCBs at the technologically complex end of the market: multilayer rigid (up to 40 layers), double-sided, HDI (blind/buried microvias, 1-mil materials), RF/mixed-dielectric, and its signature flex-rigid boards [S1]. The work is short-run and quick-turn — prototypes, pre-production, and low-to-medium-volume builds — where engineering difficulty and turnaround (days to ~2 months) matter more than price [S1]. It also occasionally imports high-volume boards from S.E. Asia as a trader, but that is immaterial today [S1]. Customers are defense/aerospace primes and subsystem makers (the dominant and growing bucket), medical-device OEMs, and industrial customers; ~109 Israeli and ~66 foreign customers in FY2025 [S1]. Sales run direct and through the Delaware sales subsidiary Eltek USA Inc. [S1].
How it is paid. Pricing is value/qualification-based, effectively cost-plus on a high-mix book — not commodity spot pricing. The whole competitive premise (Step 02/10) is that defense and medical customers will pay for a qualified, trusted, non-Chinese, high-reliability supplier. Management's pricing power is real but bounded: in the FY2025–Q1 2026 trough it tried to pass cost inflation through and met customer resistance — CEO Yaffe: "I don't have a choice… I have to load it on the prices to our customers," then "we got objection from our customers" [S3]. So pricing offsets cost slowly, not instantly.
The value chain and where the switching points sit. Upstream: laminates/copper-clad, copper foil, prepreg, and gold — several single-source suppliers, and management flags AI-driven raw-material shortages and price spikes (CCL, gold) [S1]. Internal value creation: the capital-intensive fabrication process (imaging, plating, lamination, drilling, coating, electrical test) at the single Petach-Tikva plant — the new plating/coating line is the centerpiece of the capacity program. Downstream: defense primes (Elbit, IAI, Rafael) and OEMs that design Eltek's board into a qualified system. The switching cost lives at that qualification step — once a board is designed-in and qualified to a flight or weapons program, requalifying a second source is expensive and slow, which makes repeat program orders sticky. But stickiness is per-program and informal; there are no long-term volume contracts, so demand still arrives as discrete POs and can air-pocket (exactly the Q1 2026 "late phasing" event) [S2][S3].
The unit economics that drive everything. Because the cost base (skilled labor, the plant, the machines) is largely fixed in the short run, gross margin is overwhelmingly a function of (1) how much qualified revenue runs across it — utilization — and (2) the mix (flex-rigid and high-layer-count boards carry far higher margins than "medium-tech" rigid; the Q2 2024 miss was explicitly a customer shift toward lower-margin medium-tech boards, ~42% rigid / 58% flex vs. a normal 30/70) [S3]. Then (3) labor cost and yield (the FY2025 relocation/ramp crushed yields and added training cost), and (4) FX — Eltek is NIS-functional, so when the USD weakened ~12.5% vs. the shekel in 2025 its USD-reported NIS costs inflated, a ~$2.2M margin hit independent of operations [S1][S3]. The operating leverage is the single most important quantitative fact about the model: management says incremental revenue contributes ~$0.50/$1 to gross profit [S3] — which is why $51.8M of record revenue produced collapsing profit (the incremental dollars came at depressed mix/utilization with FX and start-up drag), and why a normalized $60M at good mix could swing the P&L hard the other way.
Recurring vs. transactional. None of the revenue is contractually recurring. It is program-based and behaviorally repeating (defense programs run for years), so there is durability without contractual visibility. The closest forward indicator is backlog — but Eltek's backlog is POs within 24 months and is explicitly "not necessarily indicative," and notably the formal year-end backlog fell to $15.8M (Dec 2025) from $23.1M (Dec 2024) even as management described demand as strong [S1] (the Q1 2026 "doubling" rebuilt off that lower base via the $2.4M + $5.3M orders). This gap between qualitative demand enthusiasm and the quantitative 24-month backlog is a core tension carried into Steps 03/05/16.
5. Evidence and Sources
- End-market mix, products, capabilities, certifications, customers, facilities, raw-material/single-source and machinery risks, capacity program: 20-F FY2025 Item 4 [S1].
- Operating-leverage (~$0.50/$1), mix dynamics (medium-tech shift, rigid/flex ratio), pricing-pushback, FX magnitude: earnings transcripts Q2 2024, Q2 2025, Q1 2026 [S3].
- Margin trajectory and the three-factor margin-compression attribution: 20-F Item 5 + XBRL [S1][S2].
6. Assumption Register Updates
- A10 (new): Incremental gross margin on new revenue ≈ 50% (operating leverage) — Estimate, High sensitivity, basis = Q2 2025 management statement [S3]. This is the key swing factor for the recovery scenarios (Steps 13–15).
- Reaffirms A02 (NIS functional currency) and A07 ($60M @ 26–28% target) from Step 00.
7. Tables and Calculations
End-market mix (% of production) [S1]:
| End market | FY2023 | FY2024 | FY2025 |
|---|---|---|---|
| Defense & aerospace | 51% | 65% | 73% |
| Industrial | 14% | 13% | 9% |
| Medical | 7% | 6% | 7% |
| Distributors / CEMs / other | 28% | 16% | 11% |
Unit-economics driver map:
| Driver | Direction | Evidence |
|---|---|---|
| Capacity utilization | Higher util → fixed-cost absorption → margin up | FY2023 28% GM vs FY2025 15% / Q1'26 gross loss [S1][S2] |
| Product mix | Flex-rigid / high-layer / defense → higher margin | Q2'24 medium-tech shift cut GM to 16% [S3] |
| Labor & yield | Wage inflation + ramp/training → margin down | FY2025 COGS +21% on +11% revenue [S1] |
| FX (NIS base) | USD weakness → reported costs up | ~$2.2M FY2025 drag; −$1.3M financial result [S1][S3] |
| Operating leverage | ~$0.50 incremental GP per $1 revenue | Q2'25 call [S3] |
8. Open Questions and Data Gaps
- What is the normalized utilization/mix once the plating/coating line is live and relocation is digested? (Steps 05, 13)
- How much pricing power does Eltek truly have to pass through cost inflation given documented customer pushback? (Steps 08, 10)
- Is the gap between bullish demand commentary and the −32% YoY formal backlog a timing artifact or a demand signal? (Steps 03, 05, 16)
Next-Step Dependencies
Step 02 reuses the end-market mix and the qualification/trusted-supplier premise to place Eltek in market structure and freeze the peer set (TTMI true comp; PCBT/NCAB partial). Step 03 decomposes revenue by end-market/geography and tackles the backlog tension head-on.
Source Index
| Tag | Document | Section | Date | Notes |
|---|---|---|---|---|
| [S1] | ELTK_financials/sec_filings/20F_FY2025_summary.md |
Item 4, Item 5 | 2026-06-17 | Products, mix, geography, customers, certs, facilities, risks |
| [S2] | ELTK_financials/xbrl/xbrl_summary.md |
IS FY2009–2025 | 2026-06-17 | Margin trajectory |
| [S3] | ELTK_financials/earnings/transcript_Q2_2024.md, transcript_Q2_2025.md, transcript_Q1_2026.md |
Prepared + Q&A | 2026-06-17 | Operating leverage, mix, pricing pushback, FX |
Financial Snapshot
Step 04 — Financial Statement Quality and Adjustments
Company: Eltek Ltd (ELTK) | Date: 2026-06-17
1. Key Findings
- Reported earnings quality is high; the accounting is clean and conservative. SBC is immaterial (~$0.5M/yr, ~1% of revenue) [S2], there are no aggressive non-GAAP adjustments (management's only adjusted metric is plain-vanilla EBITDA), revenue recognition is point-in-time on shipment, and cash flow has historically tracked or exceeded net income [S2]. The earnings problem at Eltek is operational (margin), not quality (accounting).
- The FY2025 margin collapse is ~⅔ transitory and ~⅓ sticky. Management's three causes — (a) production-relocation/start-up inefficiency [genuinely one-time], (b) ~$2.2M FX drag from USD weakness [exogenous, recurring risk], (c) wage inflation [sticky] [S1][S3]. For a normalized earnings base, (a) reverses, (b) is a wash over a cycle but unhedgeable, and (c) stays — so normalized gross margin should sit below the FY2023 28% peak but well above the FY2025 15% trough.
- Do NOT anchor normalized earnings on reported FY2013 or FY2021 net income — both were inflated by large one-time tax benefits ($2.98M deferred-tax release in FY2013; $3.54M tax benefit in FY2021) [S2]. FY2021's $0.86 EPS is not a clean operating year. The cleanest "good-times" operating references are FY2023 (28.1% GM, 15.6% OM) and FY2024 (22.2% GM, 9.4% OM) [S2].
- The Nistec related-party web is disclosed and stated arm's-length, but is a structural governance discount, not a fraud flag. Sales to Nistec ($1.66M, 3.2% of revenue, +116% over 2 yrs), a $417K/yr management fee paid to Nistec, and capped services purchased from Nistec [S1] — all related-party, all approved, none hidden. Quality concern is alignment, addressed in Steps 08/17.
- Adversarial sweep: CLEAN. No short-seller reports, fraud allegations, class actions, SEC investigations, or whistleblower complaints found targeting Eltek as of 2026-06-17; short interest is negligible (~0.3%, down 33% recently) [S4][S5]. The only documented items are minor Israeli environmental penalties and historical Nasdaq listing-compliance scares (2017–2018) during the last deep trough — both resolved [S1][S6].
Net for thesis: Net positive on quality — the numbers can be trusted, and the trough is operational/transitory in large part — which is precisely what makes the recovery case investable rather than a value trap built on suspect accounting.
2. Implications for Thesis and Valuation
- Normalized earnings base for valuation (Steps 13–14): start from the operating P&L (not reported tax-distorted NI). A reasonable normalized gross margin lands in the 22–26% band (above FY2025's 15%, at/below FY2023's 28%), reflecting reversal of relocation drag but persistence of higher wages. Tax should be normalized to the ~16–23% Israeli range (Preferred Enterprise 16% election available), not the distorted single-digit/benefit years [S1][S2].
- EBITDA is a usable proxy here (clean adjustments, low SBC), but capitalize the operating lease (
$1.2M/yr → PV into debt) and treat the capacity capex cycle explicitly — FCF has been negative two years on the build, so EV/EBITDA on trough EBITDA (107x) is meaningless; normalized EBITDA ($5–6M at recovery) is the right denominator [S1][S2]. - Clean adversarial record removes the tail-risk discount a controlled micro-cap might otherwise carry — there is no Muddy-Waters-style overhang. The governance discount is purely the Nistec alignment/minority question, not integrity.
3. Objective
Convert reported numbers into an analytically usable earnings base; test "one-time" items for recurrence; analyze SBC, FX, tax, leases, related-party, and accruals; and run the mandatory adversarial sweep.
4. Narrative Analysis
Reconciling reported to usable. Eltek's statements are U.S. GAAP, audited (EY/Kost Forer since FY2024; Deloitte/Brightman through FY2023 — a routine auditor transition with both reports in the FY2025 20-F and no disclosed disagreement) [S1]. Management presents EBITDA as its headline adjusted metric, and the bridge is unremarkable: FY2025 net income $0.8M + D&A $2.1M + interest + tax ≈ EBITDA $4.5M [S2]. There is no parade of "adjusted" add-backs — no recurring "restructuring," no serial "one-time" items inflating a non-GAAP number. That alone distinguishes Eltek from many micro-caps.
Are the "one-time" margin items really one-time? This is the crux of Step 04. Management attributes the FY2025 gross-margin fall (22.2%→15.4%) to three things [S1][S3]: (1) production-relocation and new-line start-up inefficiency — genuinely transitory; once the lines are qualified and relocation is digested, under-absorption reverses; (2) ~$2.2M FX drag from the ~12.5% USD depreciation vs. NIS — exogenous and unhedgeable, a wash over a full cycle but a recurring risk, not a one-time charge; (3) wage inflation for skilled operators — sticky; this does not reverse. The honest conclusion: roughly two-thirds of the compression (relocation + the cyclical part of FX) should reverse, but a portion (wages) is permanent — so normalized margin is below the 28% peak. The Q1 2026 gross loss adds a fourth, temporary factor: an unusually low-priced shipment mix (older backlog) [S3].
SBC, tax, leases. SBC is trivially small (~$0.5–0.6M/yr) and already expensed — no dilution game [S2]. Tax is the main reported-NI distortion historically: FY2013 and FY2021 carried large one-time benefits ($2.98M, $3.54M) that make those years' EPS unrepresentative [S2]; the plant's "Benefited/Preferred Enterprise" status offers a ~16% rate option, and $10.8M of capital-loss carryforwards exist but only shelter capital gains (limited operating value) [S1]. The single 90,000 sq ft plant is an operating lease ($1.2M/yr, extended to 2039), carried as an ROU asset/liability — for valuation it should be PV'd into a debt-like obligation [S1].
Accruals / cash conversion. No red flags. FY2023 CFO ($8.9M) exceeded NI ($6.4M); FY2025 CFO ($1.1M) modestly exceeded NI ($0.8M) despite a ~$3.7M working-capital build (inventory $11.2M and AR $14.8M grew as the company stocked materials and carried receivables through the transition) [S1][S2]. The inventory build is consistent with deliberate war-driven buffering management described, not channel-stuffing (there is no channel). The one capital-allocation oddity: a $1.276M dividend paid in FY2025 that exceeded net income and pushed retained earnings slightly negative [S2] — examined in Step 07.
Adversarial sweep (mandatory) — results. Multiple independent searches (short-seller reports; fraud/accounting allegations; class actions; SEC/regulatory investigations; named activist firms Hindenburg/Muddy Waters/Spruce Point/Kerrisdale/Viceroy; whistleblower/governance controversy) returned no material adverse findings [S4][S5]. Short interest is negligible (~5,700 shares, ~0.3% of shares, down 33% as of Mar 2026) — there is no active short thesis [S5]. Documented items, all disclosed and minor: (i) Israeli environmental penalties — ~$0.6M Clean Air (2019–20, paid, 10% refunded), ~$0.1M hazardous materials (2022, partially reduced), an ongoing soil/groundwater contamination investigation, and a Feb-2026 hearing summons over two 2025 hazardous-materials incidents [S1]; (ii) historical Nasdaq listing-compliance scares — minimum-bid-price (Jan 2017) and stockholders'-equity-below-$2.5M (Oct 2018) notices during the FY2016–18 loss trough, since cured [S6]. (Namesake caution: "Eltek AS/Eltek Norway," a Delta-owned power-electronics firm, is unrelated.) This is a positive finding worth stating plainly: no integrity overhang.
5. Evidence and Sources
- Margin-compression attribution, related-party detail, tax/enterprise status, lease, environmental penalties: 20-F [S1].
- SBC, tax-benefit years, cash-flow/NI, working capital: XBRL [S2].
- FX magnitude, transition/wage commentary: transcripts [S3].
- Adversarial sweep: WebSearch results [S4][S5]; Nasdaq compliance history [S6].
6. Assumption Register Updates
- A15 (new): Normalized gross margin ≈ 22–26% (relocation drag reverses; wage inflation sticky; below FY2023 28% peak) — Judgment, High sensitivity — feeds Steps 13–15 [S1][S2][S3].
- A16 (new): Exclude FY2013/FY2021 one-time tax benefits from any normalized-earnings anchor — Fact, Medium [S2].
- A17 (new): No integrity/short-seller overhang (adversarial sweep clean) — Fact, Low [S4][S5].
7. Tables and Calculations
Reported vs. quality-adjusted read [S1][S2]:
| Item | Reported | Quality note |
|---|---|---|
| SBC | ~$0.5M/yr (~1% rev) | Immaterial; no add-back games |
| Non-GAAP metric | EBITDA only | Clean bridge (NI + D&A + int + tax) |
| FY2013 NI $3.8M | incl. $2.98M tax benefit | One-time — exclude from normal |
| FY2021 NI $5.0M | incl. $3.54M tax benefit | One-time — exclude from normal |
| Operating lease | ~$1.2M/yr to 2039 | PV into debt for valuation |
| Related-party (Nistec) | $1.66M sales / $0.68M expense | Disclosed arm's-length; alignment flag |
Margin-compression decomposition (FY2025) [S1][S3]:
| Cause | ~Nature | Reverses? |
|---|---|---|
| Relocation / start-up inefficiency | One-time | Yes (on line completion) |
| FX (~$2.2M, USD −12.5% vs NIS) | Exogenous | Cyclical wash; unhedgeable |
| Wage inflation (skilled labor) | Structural | No (sticky) |
8. Open Questions and Data Gaps
- What gross margin does the plant actually print once relocation/ramp ends — closer to 22% or 26%? (Step 05, 13)
- Will the environmental matters (Feb-2026 hazardous-materials hearing; soil/groundwater investigation) escalate beyond immaterial? (Step 11)
- Q1 2026 debt: confirm the bank-line draw behind the new net-debt position (Step 06).
Next-Step Dependencies
Step 05 reads the normalized-margin band (A15) into the quarterly momentum and KPI build. Step 06 examines the Q1 2026 debt emergence. Steps 13–14 use the normalized-earnings base and the lease-capitalization note.
Source Index
| Tag | Document | Section | Date | Notes |
|---|---|---|---|---|
| [S1] | ELTK_financials/sec_filings/20F_FY2025_summary.md |
Item 5; RP note; tax; environmental | 2026-06-17 | Margin causes, related-party, lease, penalties |
| [S2] | ELTK_financials/xbrl/xbrl_summary.md |
IS/CF; tax-benefit footnotes | 2026-06-17 | SBC, one-time tax years, CFO vs NI |
| [S3] | ELTK_financials/earnings/transcript_Q4_2025.md, transcript_Q1_2026.md |
Prepared + Q&A | 2026-06-17 | FX magnitude, transition/wage |
| [S4] | WebSearch — short/fraud/class-action/SEC + named activist firms | — | 2026-06-17 | No adverse findings |
| [S5] | MarketBeat / Nasdaq short interest | — | 2026-06-17 | ~0.3% short, down 33% |
| [S6] | WebSearch — SEC 6-K FY2017/2018 | Nasdaq compliance notices | 2026-06-17 | 2017 min-bid; 2018 equity <$2.5M (cured) |
Deeper Financial Analysis
The fundamental tier adds 9 additional research dimensions for $ELTK.