SANM · Analyze

Price at analysis: $203.35 | Mkt cap $10.90B | EV $11.68B | Sector: Technology / Electronic Manufacturing Services (EMS) — filed under Infrastructure for site taxonomy.

0. Knowledge check

python .mcp/kb.py find SANM returned no matches — no prior note, no prior report. This is a from-scratch initiation. Knowledge/Playbook/pitfall-vendor-forward-eps-is-the-wrong-fiscal-year was checked before trusting any forward multiple (see §3 — it fired).

1. General Health (Fundamentals)

TTM reconciliation — SANM's fiscal year ends ~Sept 30, so "TTM" (Jul 2025–Jun 2026) is not the same window as the FY2025 annual column (Oct 2024–Sept 2025). Summed from the four most recent quarters (2026-06-30, 2026-03-31, 2025-12-31, 2025-09-30):

Metric	TTM (sum of 4Q)	FY2025 (annual)	FY2024	FY2023
Revenue	$12.76B	$8.13B	$7.57B	$8.94B
Gross profit	$1,150.6M (9.02% GM)	$716.4M (8.81%)	$640.4M (8.46%)	$743.2M (8.31%)
Operating income	~$530.3M (7.04% OM, per Yahoo snapshot)	$395.1M (4.86%)	$345.7M (4.57%)	$461.7M (5.16%)
Net income (common)	$308.1M (2.41% NM)	$245.9M (3.02%)	$222.5M (2.94%)	$310.0M (3.47%)
Diluted EPS	$5.58 (TTM)	$4.46	$3.91	$5.18

The TTM jump is the ZT Systems acquisition (Sanmina bought AMD's ZT Systems data-center manufacturing business for up to $3B, closed 2025-10-27) plus organic AI-server demand — it isn't comparable growth, it's a step-change in the revenue base. Note the gap between operating-margin expansion (4.9% → 7.0%) and net-margin compression (3.0% → 2.4%) on a TTM basis: this is the acquisition showing up below the operating line — TTM interest expense $94.5M (vs $5.2M FY2025, because total debt jumped from $300M to $2.17B to fund the deal), plus ~$170M of TTM restructuring/M&A "unusual items," plus minority-interest drag. This is an integration-cost story, not an earnings-quality red flag — but it means net margin understates the underlying operating improvement while the deal digests.

Free cash flow. TTM OCF $901.1M − Capex $306.9M = FCF $594.1M (FCF/share ≈ $10.80–11.08). Data trap: Yahoo's get_stock_info snapshot reports freeCashflow: -$222.2M for the same TTM window — this directly contradicts OCF/Capex summed from the same vendor's own quarterly cash-flow statements. The quarterly-level Free Cash Flow field sums correctly to $594.1M; only the pre-computed snapshot aggregate is wrong, likely mis-netting the ZT purchase-of-business investing outflow into "free cash flow." Always recompute FCF from the quarterly cash-flow statement on this name, especially around the acquisition quarter — do not quote the snapshot field. Flagging as a new pitfall candidate (see block below).

Balance sheet — leveraged up materially for the deal. - Total debt: $300M (Sep-2025) → $2.17B (Jun-2026); net debt only $327M given $1.84B cash. - Debt/Assets: 5.1% (pre-deal) → 22.3% (post-deal). D/E ≈ 85–88%. - Interest coverage: TTM EBIT (as-reported) ≈ $552M ÷ TTM interest expense $94.5M ≈ 5.85x — adequate, not stretched. - Inventory ballooned from $1.99B (Sep-2025) to $3.15B (Jun-2026), +65% — working-capital build to support the AI-server ramp. In a capital-intensive, thin-margin business this cuts both ways: it supports near-term revenue visibility but is a direct exposure if AI-capex demand air-pockets. - Current ratio 1.78, quick ratio 0.99 — inventory-heavy, as expected for EMS.

Shares outstanding: 57.4M (FY2022) → 53.6M (latest) — a genuine buyback-driven -4.4% 3yr CAGR, though buybacks paused this year to conserve cash for the ZT deal (a Trefis piece flagged this: "Sanmina Stopped Retiring Stock, And Then Its Shares Pulled Back").

Per-share: Revenue/share ≈ $147 (FY2025 annual basis) / TTM revenue/share materially higher given the deal; FCF/share ≈ $10.80–11.08 TTM.

2. Moat & Competitive Advantage

Quantitative base. ROIC is modest: TTM NOPAT (EBIT ~$530M × ~73% after-tax) ≈ $387M against average invested capital that moved from $2.65B to $4.73B over the year (deal-driven) — call it ROIC ≈ 8–10%, roughly at or just above cost of capital. Gross margin has crept up slowly over four years (7.85% → 8.31% → 8.46% → 8.81% → 9.02% TTM) as mix shifts toward higher-complexity cloud/AI work, but this remains a thin-margin, capital-intensive business by construction — judge it on capital discipline and cycle survival, not SaaS-style margin structure.

Adversarial stress-test. EMS has essentially no network effects, weak brand power, and modest switching costs (qualified manufacturing lines, ramped capacity, and regulatory quals in defense/medical/aerospace work create some friction, but a well-funded rival — Celestica, Jabil, Flex, or Foxconn/Wiwynn on the ODM side — could contest any given program over a 2–4 year horizon). The single biggest business-model fact this cycle: the growth is not diversified. Per the FY2025 10-K, the top-10 customers were ~52% of sales; post-ZT, reporting puts top-10 concentration at ~65% in Q1 FY26, and the ZT Systems book itself is ~100% AMD. Sanmina did not buy a diversified AI franchise — it bought a large, high-growth, single-customer contract manufacturing relationship. AMD itself now holds ~2.1% of SANM (received as deal consideration), which aligns incentives somewhat but also deepens the entanglement.

Revenue-stream map. Two segments: Integrated Manufacturing Solutions (IMS, now dominated by cloud/AI infrastructure + ZT/AMD) and Components/Products/Services (higher-margin, more diversified — PCBs, backplanes, cable assemblies, optical/RF, defense/aerospace repair). The legacy diversified book (industrial, medical, defense/aerospace, comms) is real and durable but is now a shrinking share of the mix as AI/cloud dominates growth.

Disruption forecast. Two vectors: (1) hyperscalers/AMD could diversify manufacturing across multiple EMS partners or bring more in-house over a multi-year horizon — normal EMS-industry risk, sharpened here by concentration; (2) an AI-capex deceleration would hit a business that just levered up and built inventory to chase this specific demand curve — the operating-leverage exposure is now larger than it was 18 months ago.

Evergreen assessment. EMS as an industry is evergreen (someone will always contract-manufacture electronics), but Sanmina's current earnings power is not evenly evergreen — a large slice of the TTM step-change rides one customer relationship inside one secular trend. Moat rating: weak structural moat, real but narrow cyclical tailwind.

3. Valuation

Pitfall fired: pitfall-vendor-forward-eps-is-the-wrong-fiscal-year. Yahoo's headline forwardPE = 14.61x on forwardEps = $13.92 is FY2027 (fiscal year ends ~Sept), not the current fiscal year. The correct current-year (FY2026, ending Sept-2026) read is priceEpsCurrentYear = 16.81x on epsCurrentYear = $12.10 — which matches the company's own raised guidance from the 2026-07-27 Q3 print almost exactly (non-GAAP EPS guide $11.90–$12.20, midpoint $12.05, on FY2026 revenue guide $14.0–14.3B). Trailing P/E (36.4x on TTM EPS $5.58) is a third, even less useful number here: it is mechanically distorted downward-lagging because the TTM window still carries the pre-ZT Q4-FY25 quarter (EPS $0.88); as that quarter rolls off and Q4-FY26 (~$7.3 implied) rolls in over the next two quarters, trailing EPS will converge rapidly toward the ~$12 current-year run-rate. Do not anchor on the 36.4x trailing multiple as a valuation signal — it is a transitional artifact, not a "expensive" reading.

No dividend (yield 0%, payout 0%) — DDM/DYT are N/A.

Graham IV = √(22.5 × 5.58 × 47.76) = $77.4 vs price $203.35. Graham's formula assumes stable, book-anchored earnings; it is not a useful anchor for a cyclical EMS name mid-re-rating on a growth narrative (P/B is already 4.3x, itself rich for EMS, which historically trades 1–2x book) — noted for completeness, not weighted in the verdict.

Peer read (informational, not independently verified line-by-line): Jabil trades near/above 20x forward on its AI-infrastructure-led guide raise; Celestica has re-rated to a premium multiple riding the same AI-server wave. On a like-for-like current-fiscal-year basis, SANM's 16.8x looks statistically cheap next to both — but the concentration risk (single-customer-driven vs. Jabil's broader diversified base, Celestica's own hyperscaler mix) is the reason for some of that gap, not pure mispricing.

Fair value range: 14–18x FY2026 consensus/guide EPS (~$12.05) = $169–$217; cross-checked against 13–16x FY2027 consensus EPS (~$13.92, itself only 15% above FY26 — a plausible normalization after the initial ramp) = $181–$223. Blending: $170–$220.

Current price $203.35 sits in the upper half of fair value — not a clear bargain, not stretched. Down ~30% from the 52-week high ($288.68, following a 2026-07-28 JPMorgan downgrade to Neutral, price target cut $275→$260), the market is already pricing in some concentration/integration skepticism.

4. Type-Specific Overlay

Not a dividend grower, REIT, or BDC. Market cap $10.9B is mid-cap, not small/micro — the §6 small-cap framework does not apply. The relevant overlay is the general EMS lens already applied throughout §1–3: judge on ROIC, working-capital discipline, customer concentration, and FCF through a cycle, not growth-SaaS multiples.

5. Sentiment

Q3 FY2026 (reported 2026-07-27): revenue $3.46B, +69.7% YoY; IMS segment +79.4% YoY; ZT Systems contributed ~$1.1B of the quarter. Non-GAAP diluted EPS $3.31, +116% YoY, non-GAAP operating margin 8.0%. Management raised FY2026 guidance to $14.0–14.3B revenue / $11.90–12.20 non-GAAP EPS. Despite the beat-and-raise, the stock is down materially from its 52-week high and JPMorgan moved to Neutral with a lowered price target — the sell-side reaction reads as "good quarter, but the multiple/concentration risk is now the debate," consistent with the valuation read above.

Insider activity is one-sided. Since late 2024 there is a long run of officer/director sales (CEO Sola sold 118,368 sh at $228.84 on 2026-05-06; CFO Faust and EVP Reid sold repeatedly through Feb–May 2026 as the stock ran from ~$75 to $200+) against essentially one small insider purchase (a director, $99.7K at $142.46, 2026-01-30). This is largely explainable as routine profit-taking/ 10b5-1 selling into a huge run rather than a distinct red flag, but there is no offsetting insider buying signal to lean on. Institutional ownership is very high (95%), led by FMR/Fidelity (15.0%), BlackRock (10.25%), Vanguard entities (~10% combined) — AMD itself holds 2.15% (deal consideration).

6. Synthesis & Verdict

Two questions, per framework: 1. Is this a good business? Mixed. Improving structural gross margin, real (if narrow) moat in complex/regulated EMS niches, disciplined balance sheet historically — but ROIC is only modest (~8–10%), and the current earnings step-change is concentrated in one customer relationship (AMD/ZT) riding one secular cycle (AI datacenter capex), funded by a large debt-financed acquisition and a working-capital build. This is a real tailwind, not a proven moat. 2. Has the market priced it in? Roughly — fair value $170–220 vs price $203.35 puts SANM in the upper-middle of range: not a clear value case, not obviously overpriced either, especially against richer-multiple peers (Jabil, Celestica) on a like-for-like current-year basis.

Tension worth naming explicitly: Fundamentals/Valuation read "statistically reasonable, even cheap vs. peers" on the current-year multiple; Moat reads "concentration and leverage risk that peers with more diversified AI exposure don't carry to the same degree." Weighted for a capital-intensive cyclical EMS name, concentration + leverage risk is not disqualifying — the company reports strong FCF and adequate interest coverage even after the deal — but it is enough to keep conviction moderate rather than high, and to prefer entering on weakness rather than at the current price.

Verdict: WATCH. Conviction 5.5/10. Fair value $170–220. Entry $150–175 (a pullback that compensates for the AMD/ZT concentration and post-acquisition leverage). Trim 38x ttm (~$212) — set on trailing EPS per the pitfall note's standing guidance ("set watchlist trims on ttm rather than fwd whenever fwd/current-year disagree"); this trailing multiple looks high in isolation only because trailing EPS is still catching up to the ~$12 FY26 run-rate — expect the dollar level to stay roughly stable while the multiple compresses mechanically as low legacy quarters roll off over the next two prints.

Key risks: - Customer concentration — top-10 ~65% of sales post-ZT, ZT book ~100% AMD. A single contract-repricing or AMD-diversification decision has an outsized effect on this name relative to EMS peers. - Leveraged into a cyclical bet — debt tripled to fund the deal, inventory up 65% YoY to chase AI-server demand; an AI-capex deceleration hits a thin-margin, now-more-levered balance sheet harder than it would have 18 months ago. - Transitional trailing multiple — the 36.4x trailing P/E is not a valid "expensive" signal right now; don't let a future headline trailing-multiple compression (as legacy quarters roll off) be mistaken for a re-rating catalyst on its own.

Data traps hit

1. pitfall-vendor-forward-eps-is-the-wrong-fiscal-year — FIRED. forwardPE 14.61x (forwardEps $13.92) is FY2027; the true current-year (FY2026) multiple is priceEpsCurrentYear 16.81x on epsCurrentYear $12.10, which matches company guidance almost exactly.
2. New candidate trap — Yahoo snapshot freeCashflow disagrees with its own quarterly statements. get_stock_info reports TTM freeCashflow: -$222.2M; summing the same vendor's quarterly Free Cash Flow field (OCF − Capex) over the same four quarters gives +$594.1M. The mismatch appears specifically around the ZT Systems acquisition quarter (large purchase-of-business investing outflow); the snapshot aggregate likely mis-nets it into "free cash flow." Recompute FCF from the quarterly cash-flow statement whenever a large M&A quarter sits inside the TTM window — don't trust the snapshot field.

---

Sources: python .mcp/fin.py SANM --news; Yahoo Finance MCP (get_stock_info, get_financial_statement quarterly income/cashflow/balance sheet, get_holder_info insider + institutional, get_recommendations); WebSearch (Q3 FY26 earnings coverage, ZT Systems deal terms, FY2025 10-K customer-concentration language, EMS peer positioning). roic.ai quarterly/TTM ratios were unavailable on the connected free tier (annual-only) — not used.