Phase A — Understand the business
Lens 1 · Company Overview
What it is. Jabil is one of the world's largest providers of engineering, manufacturing, and supply-chain solutions — the contract-manufacturing (EMS / "manufacturing solutions") layer that companies outsource to when they want someone else to design-for-manufacturability, procure components, and run high-volume global production. FY2025 net revenue $29.8B, net income attributable to Jabil $657M, ~135,000 employees across ~100 sites in 30 countries, ~35M sq ft of factory floor (14M owned / 21M leased). HQ St. Petersburg, FL; NYSE: JBL; Delaware-incorporated; FYE August 31.
How it makes money. "Turnkey" manufacturing: Jabil buys the components, builds the product to customer spec, and ships it. Revenue per product = (a) a materials pass-through element + (b) a labor/overhead element; gross margin lives almost entirely in (b), which is why blended GM is structurally thin (8.9% FY2025 ). It commits significant working capital to procurement (inventories $4.7B, AR $4.0B) and recovers it through scale, utilization, and inventory turns. The business is run through dedicated, customer-facing business units — a single point of contact per customer — backed by shared global procurement.
Three reportable segments (re-cut as of Sept 1, 2024):
- Intelligent Infrastructure — the AI engine: cloud & data-center infrastructure, capital equipment (incl. semicap), networking & communications. 41% of FY2025 revenue, the growth driver.
- Regulated Industries — automotive & transportation, healthcare & packaging, renewables & energy infrastructure. 40% of FY2025 revenue, the ballast (stickier, qualification-gated, higher-margin).
- Connected Living & Digital Commerce — digitalization/automation incl. warehouse automation & robotics, plus the residual consumer/connected-living and digital-commerce business. 19% of FY2025 revenue, shrinking (post-Mobility-divestiture).
Customers. Top-5 customers ≈ 36% of FY2025 revenue; 87 customers ≈ 90% of revenue. The standout: Customer A = 16% of FY2025 revenue, reported primarily in Intelligent Infrastructure — i.e. a single hyperscaler/cloud customer is now Jabil's largest relationship by a wide margin. The prior dominant customer ("Customer B," 17% FY2023 / 11% FY2024, in Connected Living) faded as the Mobility/consumer business was divested.
Verdict on the model: a low-margin, high-working-capital, scale-and-utilization business that has deliberately repositioned its mix toward the highest-demand end-market in the economy (AI data centers) while buying its way up the value chain (cooling, power). The model itself is commodity; the mix shift and capital allocation are the story.
Lens 2 · Supply Chain
Jabil sits in the middle of the electronics value chain — it is itself a supplier-of-manufacturing to brand owners, and a buyer from component makers. Map, with named stakeholders:
Upstream (Jabil buys):
- Silicon / compute: Nvidia GPUs, AMD, Broadcom ASICs, and the memory/networking silicon that goes onto the boards and systems Jabil assembles for data-center customers (Jabil integrates, it does not fab).
- Components & passives: procured "assembly-by-assembly" from a broad global supplier base via regional sourcing hubs; some components are single-source and periodically allocated — supply shortages can curtail production of an entire assembly. This is the named chokepoint in Jabil's own disclosure.
- Thermal / cooling IP: now in-housed via the Mikros Technologies acquisition (Oct 2024) — liquid-cooling cold plates and thermal management for high-density GPU racks.
- Power / energy infrastructure: in-housed via Hanley Energy — data-center power systems (named by management as a higher-margin contributor lifting FY2027 margins).
The company (Jabil): design-for-manufacturability → component procurement → automated/continuous-flow assembly → system integration → test → logistics/fulfillment, run across China, Malaysia, Mexico, and the US. 75% of revenue is foreign-sourced (down from 86% in FY2023 as US data-center work reshored).
Downstream (Jabil sells to):
- Hyperscalers / cloud — the Intelligent Infrastructure buyers; "Customer A" (16%) is here. The AI data-center buildout (Amazon, Microsoft, Google, Meta, plus neoclouds and Nvidia's ecosystem) is the demand source.
- Capital-equipment / semicap OEMs, networking & comms OEMs.
- Regulated OEMs: automakers/Tier-1s, medical-device and pharma companies (the latter now served deeper via the Pharmaceutics International CDMO acquisition), renewables/energy-infrastructure customers.
- Consumer/connected-living and digital-commerce brand owners (the shrinking tail).
Chokepoints & dependencies: (1) single-source components (Jabil's own flagged risk); (2) GPU/accelerator allocation upstream — Jabil's AI revenue is gated by how many accelerators its hyperscaler customers can get; (3) customer concentration downstream — one 16% customer; (4) geographic concentration in China/Mexico exposes it to tariff/trade policy. Names present — this is not a generic chain.
Lens 3 · Competitive Advantages (moats)
Be honest: EMS is a structurally low-moat industry. Jabil competes on cost, time-to-market, scale, global footprint, and qualification — and it competes against its own customers' option to in-source. The principal competitive factors Jabil itself lists are cost, speed, efficiency, global locations, and component pricing — none of which is a durable moat in isolation. So where is the edge?
- Scale + footprint (real but shallow): ~$30B of purchasing power, 100 sites, the ability to let a customer build "simultaneously in optimal locations." This is a barrier to sub-scale entrants, not to Foxconn or Flex.
- Switching costs (moderate, rising in Regulated/AI): in regulated end-markets (medical, auto, pharma via Pii), Jabil's lines are qualified into the customer's regulatory filings — ripping that out is expensive and slow. In AI data-center, being designed into a hyperscaler's reference rack with custom liquid-cooling creates stickiness that the old smartphone-casing business never had. This is the deliberate moat-deepening trade.
- Vertical integration into the bottleneck (the genuine edge being built): Mikros (cooling) + Hanley (power) move Jabil from "we assemble your box" to "we own the thermal and power subsystems that are the actual supply constraint in an AI rack." Liquid cooling is mandatory above ~the power densities of modern GPUs; owning that IP is worth more margin than bending metal.
- Bargaining power: WEAK over hyperscaler customers (one buys 16% — they have the leverage), MODERATE over the long tail, MIXED over suppliers (single-source components mean suppliers sometimes hold the cards). Jabil is a price-taker on materials and largely a margin-taker from its biggest customers.
Moat verdict: Jabil is buying its way from a no-moat commodity assembler toward a narrow-moat supplier of supply-constrained AI subsystems. The moat is real but young, and it does not yet protect the 16%-customer relationship. Ground: positioning.md/bottlenecks.md (robotics wiki) frame the AI-hardware bottleneck as cooling/power-density — exactly where Jabil is integrating.
Lens 4 · Segments
Revenue mix by segment (% of net revenue) — every figure:
| Segment | FY2023 | FY2024 | FY2025 | Q2-FY26 YoY | 1H-FY26 YoY |
|---|
| Intelligent Infrastructure | 32% | 32% | 41% | +52% | +53% |
| Regulated Industries | 38% | 42% | 40% | +10% | +7% |
| Connected Living & Digital Commerce | 30% | 26% | 19% | −8% | −10% |
Total revenue: FY2023 $34.7B → FY2024 $28.9B (−16.8%, the Mobility divestiture) → FY2025 $29.8B (+3.2%) → FY2026 guided ~$35B. The shape is a V: revenue fell as Jabil shed ~$10B of low-margin Mobility/consumer work, then re-accelerated as AI infrastructure took over.
The driver, decomposed:
- Intelligent Infrastructure +34% in FY2025, then +52% in Q2-FY26 — almost entirely cloud & data-center infrastructure from existing customers (FY25 +30%, Q2-FY26 +42%). Capital equipment a secondary tailwind (+10% FY25). Networking has been a drag (−6% FY25) but turned slightly positive recently.
- Regulated Industries roughly flat-to-up — auto/transport soft (−2% FY25) then recovering (+6% Q2-FY26 auto), renewables/energy up, healthcare ~flat. The ballast is doing its job: low growth, but stable and margin-accretive (and the Pii acquisition adds pharma CDMO revenue here).
- Connected Living down 25% FY25 / −8% Q2-FY26 — the Mobility divestiture (−27% FY25) is the cause; digital-commerce up slightly (+2–5%) is the silver lining.
Segment economics: the 10-K markdown did not expose the Note-14 per-segment operating-income dollars (table collapsed in ingestion) — so segment OI margin = n/a at the dollar level. What is sourced: blended gross margin compressed to 8.9% (FY25) from 9.3% (FY24) on mix — Intelligent Infrastructure and Connected Living carry lower GM than the corporate average, so the very growth that lifts revenue dilutes gross margin even as core operating margin rises (5.4% FY25 → 5.7–5.8% guided FY26 → >6% guided FY27) on operating leverage and the higher-margin Mikros/Hanley/Pii mix.
Phase B — Measure performance
Lens 5 · Earnings Result
Latest print — Q3 FY2026, reported 2026-06-17:
- Revenue $8.8B (vs ~$8.66B consensus) — beat, up double-digits YoY.
- Core operating income $504M; core diluted EPS $3.16 (vs ~$3.12 consensus) — beat.
- Management raised FY2026 to: revenue ~$35B, core operating margin ~5.8%, core diluted EPS ~$12.70, and AI-related revenue ~$13.6B. Guided FY2027 core operating margin above 6%, helped by mix, leverage, and integration of higher-margin acquisitions (Hanley).
This is the third consecutive raise of FY26: AI revenue guidance walked from ~$9B → $13.1B (at Q2) → $13.6B (at Q3); core EPS from ~$11.55 → $12.25 → $12.70. The trajectory itself is the signal — Jabil keeps under-promising and the AI book keeps over-delivering.
Prior on-shelf print — Q2 FY2026 (quarter ended 2026-02-28):
- Revenue $8,282M (+23% YoY vs $6,728M); gross profit $746M (GM 9.0%); operating income $374M; net income to Jabil $223M; GAAP diluted EPS $2.08; core diluted EPS $2.69 (vs ~$2.51 consensus — beat).
- 1H-FY26: revenue $16,587M (+21%); GAAP diluted EPS $3.43; core diluted EPS $5.55 (vs $3.94 in 1H-FY25, +41%). Core OI 1H $890M (margin ~5.4%).
- Intelligent Infrastructure +52% (cloud/DC existing customers +42%); Regulated +10%; Connected Living −8%.
Full-year FY2025 anchor:
- Revenue $29,802M; gross profit $2,646M (GM 8.9%); GAAP operating income $1,182M (depressed by $181M restructuring + $53M divestiture loss); net income $657M; GAAP diluted EPS $5.92.
- Core operating income $1,620M (core OM 5.4%); core diluted EPS $9.75 (+15% vs $8.49 FY24) — the +15% despite a shrinking revenue base is the buyback at work (diluted shares 110.9M vs 124.3M).
- Adjusted FCF $1,318M (vs $1,055M FY24); CFO $1,640M, capex $468M.
Balance-sheet flags:
- Cash $1,933M; total debt $2,885M ($499M current + $2,386M LT) → net debt ~$952M; net-debt/core-EBITDA well under 1x — comfortably investment-grade.
- Equity is razor-thin at $1,517M against $18.5B assets — not distress, but the mechanical result of $7.9B of cumulative treasury stock from a decade of aggressive buybacks. Book value is not the way to value this company.
- Working-capital watch-items: AR grew $504M and inventory $431M in FY25 (a use of cash) while accounts payable + accrued jumped +$1,426M (a source) — the AP swing is the single biggest contributor to FY25 CFO. Days-payable rose on "customer-controlled consignment components." If that consignment/AP tailwind reverses, reported CFO would look materially worse than earnings. Flag.
Market reaction: despite the Q3 beat-and-raise, the stock is ~$365 (June 26), below its $385.63 ATH (June 15) — i.e. the print was good but the stock had already run ~80% in a year and is consolidating. Expectations are now demanding.
Lens 6 · Earnings Calls (sentiment trend)
No transcripts on the shelf (transcripts/ empty) — sentiment reconstructed from print cadence + management commentary. The arc across the last ~4 calls (Q4-FY25 → Q1 → Q2 → Q3-FY26):
- Tone: progressively more confident, specifically on AI. Each quarter management raised the AI revenue number and framed the back half as stronger. By Q3 FY26 the language is "AI infrastructure demand remained extremely strong" with FY27 margin >6% guidance — a forward, conviction-laden posture rather than hedged.
- What they keep saying: "core" margin expansion, operating leverage, free-cash-flow generation, disciplined capital return, "diversified" customer base (a tell — they say it because concentration rose).
- What they stopped saying: Mobility, smartphones, "consumer" as a growth vector — the consumer story has been deliberately retired from the narrative. The pivot is complete in the messaging, not just the numbers.
- New vocabulary: liquid cooling, thermal management, power (Hanley), AI factories, semicap, "Intelligent Infrastructure." The lexicon has migrated wholesale to the AI-data-center stack.
Sentiment verdict: bullish and tightening — a management team that has found its growth narrative and is leaning into it. The risk in this is asymmetry: a single soft AI quarter would be jarring against this tone.
Lens 7 · Comps
EMS / contract-manufacturing peers (the 10-K's own named peer group: Celestica, Flex, Hon Hai/Foxconn, Plexus, Sanmina). Market data as of June 26, 2026:
| Company | Ticker | Mkt cap (USD) | Fwd P/E | EV/EBITDA | Notes |
|---|
| Jabil | JBL | ~$37.8B | ~28.7x (=$365 / $12.70 core FY26 ) | n/a | The subject |
| Celestica | CLS | $41.5B | 33.7x | 28.8x | Purest AI-DC EMS comp; richest multiple |
| Flex | FLEX | ~$30B+ | 31.7x | 28.9x | Diversified, similar AI tilt |
| Sanmina | SANM | $13.0B | 20.6x | 18.2x | Cheapest; smaller, less AI-levered |
| Hon Hai (Foxconn) | 2317.TW | (Taiwan-listed) | n/a | n/a | Largest by far; Apple/AI server scale |
| Plexus | PLXS | n/a | n/a | n/a | Smaller, regulated-heavy |
Read: the entire EMS group has re-rated from its historical ~10–13x forward P/E to high-20s/low-30s on the AI-infrastructure narrative. Within that group Jabil is the cheaper large-cap AI-EMS name — ~29x vs Celestica 34x / Flex 32x — while compounding core EPS ~15% (FY25) accelerating to ~20%+ (FY26). Only Sanmina (21x) is clearly cheaper, and it is smaller and less AI-exposed. So Jabil is neither the cheap value play (that's Sanmina) nor the priciest momentum name (Celestica) — it's the GARP slot in the EMS AI trade. The whole sector multiple is the risk: if the AI-capex narrative cools, all of these de-rate toward their teens together. (Dividend yield negligible — Jabil's capital return is ~96% buyback; 5-yr ROE is distorted by the thin buyback-shrunk equity base, so a clean cross-peer ROE comp is n/a.)
Lens 8 · Stock-Price Catalysts
What has moved JBL >5% over the last ~3 years, and what it reveals:
- Aug 2023 — Mobility divestiture announced ($2.2B sale to BYD Electronic, completed late 2023): the strategic pivot. Reframed Jabil from a low-margin Apple/consumer supplier to a higher-mix industrial/AI name.
- FY2024–2025 — serial earnings beats + buyback: each quarter that beat and raised pushed the stock; the $2.5B FY24 + $1.0B FY25 buybacks mechanically lifted EPS and signalled confidence.
- Early Mar 2026 — Q1 FY26 print: stock +~10% intraday on a data-center-driven revenue beat ($6.9B vs $6.6B). Confirmed the market now trades JBL as an AI name.
- Mar 18 2026 (Q2) and June 17 2026 (Q3) — beat-and-raise: drove the run toward the $385 ATH (June 15 2026).
- Analyst-target step-ups (June 2026): Goldman → $482, Stifel → $460, JPMorgan/Raymond James → $450, Baird → $440, UBS → $430; consensus PT ~$453.67. The sell side is chasing.
Pattern: JBL now reacts almost exclusively to (1) AI/data-center revenue trajectory and guidance, and (2) capital-return signals. It has become a high-beta AI-capex proxy — it will trade with the hyperscaler capex narrative (up on Nvidia/Microsoft/Amazon capex raises, down on any "AI digestion" scare) far more than on its own diversified-industrial roots. That is both the opportunity and the trap.
Phase C — Judge people & books
Lens 9 · Management
- CEO: Michael (Mike) Dastoor — CEO since May 2024; previously CFO from 2018; ~24+ years at Jabil across global roles. A finance-trained operator now running the company. Track record: as CFO he architected the capital-return machine and the Mobility exit; as CEO he has delivered the AI pivot with serial beat-and-raises and disciplined margin guidance (5.4%→5.8%→>6%). Archetype: professional manager / capital allocator, not founder — appropriate for a scale EMS at this stage.
- CFO: Gregory B. Hebard — Principal Financial Officer.
- Chairman transition: Mark T. Mondello — 33 years at Jabil, CEO 2013–2023, executive chairman thereafter — steps off the board at the January 2026 annual meeting (along with two other directors). This is a clean generational handoff: the long-tenured CEO who ran the diversification era is fully exiting governance, leaving Dastoor unambiguously in charge. Read it as completion of a planned succession, not a rupture.
- Capital-allocation history (the standout): over FY2016–FY2025 Jabil returned $7,083M — $6,579M of buybacks + $504M dividends — and cut diluted share count from ~135.9M (FY23) to ~105.5M (now), ~22% fewer shares in three years. FY25 alone: $1.0B buyback + $36M dividend; a fresh $1.0B 2026 authorization ($865M remaining). This is a management team that treats the share count as the primary lever and has used the Mobility cash + FCF to shrink the float into rising EPS. This is the single most important fact about the equity.
- M&A discipline: three bolt-ons, all on-strategy and value-chain-accretive — Mikros (liquid cooling), Pharmaceutics International / Pii (pharma CDMO), Hanley Energy (data-center power). Small, tuck-in, margin-accretive — not empire-building.
- Red flags: (1) Insider share sales have been flagged by third-party analysts during the run-up — worth monitoring but common after an 80% move and not by itself damning (
insider-transactions.csv absent — not sourced at the transaction level). (2) Heavy reliance on non-GAAP "core" metrics — the GAAP-to-core gap is large (FY25 GAAP EPS $5.92 vs core $9.75) and management narrates on core; legitimate (restructuring, intangible amort, divestiture noise) but demands scrutiny (see Lens 10). (3) The "diversified customer base" messaging sits uneasily next to a 16% customer.
Management verdict: a credible, financially disciplined operator with an exceptional capital-return record and a clean succession — the kind of stewardship that compounds a low-margin business. The watch-item is that the same buyback that flatters EPS also leaves almost no equity cushion and depends on FCF that is partly working-capital-financed.
Lens 10 · Forensic Red Flags
Acting as a forensic analyst across the statements:
- GAAP-vs-core gap (the headline item): FY25 adjustments added +$438M to operating income ($1,182M GAAP → $1,620M core) and core EPS ($9.75) is 65% above GAAP EPS ($5.92). The add-backs — restructuring ($181M), intangible amortization ($62M), SBC ($107M), divestiture loss ($53M) — are individually defensible, but SBC ($107M, ~16% of net income) is a real, recurring, cash-dilutive cost that core EPS adds back. Value the company on core, but haircut for SBC.
- Cash flow vs earnings: FY25 CFO ($1,640M) comfortably exceeds net income ($657M) — good — but the quality is flattered by a +$1,426M accounts-payable/accrued swing driven by customer-controlled consignment components and payment timing. AR (+$504M use) and inventory (+$431M use) ran against cash. If the AP/consignment dynamic normalizes, CFO could fall below earnings. This is the metric to watch each quarter.
- Receivables/inventory vs revenue: AR rose to $4,039M (from $3,533M) and inventory to $4,681M — broadly tracking the revenue re-acceleration, not obviously outrunning it; days-in-AR and days-in-inventory actually fell in Q4 FY25 on better working-capital management. No channel-stuffing signature visible.
- Receivables-financing programs: Jabil runs trade-AR sales programs + a global asset-backed securitization program, and "other expense" rose in FY25 on higher utilization of these. This is off-balance-sheet-style liquidity that makes reported AR and CFO look better than the underlying — standard for large EMS, but it means the true cash-conversion cycle is partly financed. Flag for quality-of-earnings.
- Goodwill/intangibles: modest and rising on acquisitions (goodwill $841M, intangibles $273M) — small relative to $18.5B assets; impairment risk low.
- Thin equity / heavy treasury stock: $7.9B treasury vs $1.5B total equity — not an accounting red flag (it's deliberate buyback), but it means leverage and coverage ratios, not book-based ratios, are the right lens.
- Segment-reporting opacity: segment was re-cut Sept 2024 and per-segment operating-income dollars are not cleanly disclosed in the ingested filing — reduces the ability to verify where the margin actually sits. n/a at segment-OI level.
Regulatory findings:
- SEC Litigation Releases: none naming Jabil since 2021-06-29 (EDGAR EFTS, forms=LR). Zero findings.
- SEC AAERs: none naming Jabil since 2021-06-29 (EDGAR EFTS, forms=AAER). Zero findings.
- 10-K Item 3 (Legal Proceedings): points to Note 19 — Commitments and Contingencies; the ingested filing exposes only routine ordinary-course language, no material litigation disclosed.
- Non-SEC enforcement (web): no material FTC/DOJ/FDA/CFPB enforcement actions surfaced for Jabil Inc. in search.
- Net: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-29. Historical note: Jabil settled an FCPA matter with the SEC in 2016 (legacy, fully resolved, outside the search window) — clean since.
Phase D — Project & stress-test
Lens 11 · Forward Projection
Anchor = FY2026 management guidance (core EPS ~$12.70, revenue ~$35B, core OM ~5.8%), with diluted shares ~105.5M and falling on the active $865M buyback. Building base / bull / bear core-EPS paths for FY2026 → FY2028 (Jabil FYE August). All output, inputs labeled.
FY2026 (year nearly complete, 3 of 4 quarters reported): core EPS ≈ $12.70 — treat as near-actual.
Base case (FY27 / FY28):
- Revenue: $35B (FY26) → ~$39B (FY27, +11%) → ~$43B (FY28, +10%).
- Core operating margin: 5.8% (FY26) → ~6.2% (FY27) → ~6.4% (FY28).
- Core OI: ~$2.4B (FY27) → ~$2.75B (FY28).
- Buyback: ~$1.0B/yr retiring ~2–3% of shares annually → diluted shares ~105.5M → ~103M (FY27) → ~100M (FY28).
- Base core EPS: FY27 ≈ $15.0; FY28 ≈ $17.7.
Bull case: AI-DC demand stays vertical, networking inflects, margin reaches ~6.6% FY27 on cooling/power mix, $1.5B/yr buyback → FY27 core EPS ~$16.5, FY28 ~$20.
Bear case: hyperscaler AI-capex digestion hits in FY27 — Intelligent Infrastructure decelerates to +mid-single-digits or flattens, new-plant fixed costs drag margin back to ~5.5%, buyback continues but can't offset → FY27 core EPS ~$12.5–13 (flat), FY28 ~$13. In this world the multiple also compresses (double hit).
Valuation frame at ~$365: ~28.7x FY26 core EPS, ~24x base FY27, ~21x base FY28. The market is paying a high-20s multiple for a ~17% core-EPS grower with cyclical AI-capex exposure — priced for continued execution, not for a digestion air-pocket. A re-rate to the mid-20s is plausible if growth holds; a de-rate to the mid-teens is the downside if AI-capex stutters.
(Per --watchlist rule, no forecast.ts create logged — base case noted for a future human-gated thesis pass. If committed: "JBL FY27 (Aug-2027) non-GAAP core EPS ≥ $15.0," p≈0.55.)
Lens 12 · Bull vs Bear
Bull case. Jabil is the GARP way to own the AI-data-center buildout without GPU/cycle risk at the silicon layer. It is structurally levered to hyperscaler capex (the most-funded capex cycle in history), it is moving up the value chain into the two genuine supply bottlenecks — liquid cooling (Mikros) and power (Hanley) — which carry better margins than assembly, and it returns ~all of its growing FCF as buybacks, shrinking the share count ~2–3%/yr into rising EPS. Core EPS compounds ~15% → ~20%, core margin marches 5.4% → >6%, and at ~29x it trades below its closest AI-EMS peers (Celestica 34x, Flex 32x). Management is disciplined, the succession is clean, and the regulated-industries ballast (healthcare/auto/energy, now deepened by Pii) cushions a consumer-electronics-style downturn. The contrarian read: the market still partly prices JBL as a sleepy low-margin EMS commodity, when its mix is now majority high-growth infrastructure — there is room for the multiple to hold even as EPS compounds.
Bear case (permanent-impairment risks). (1) Customer concentration: one customer is 16% of revenue (in the AI segment) and the top-5 are 36% — a single hyperscaler insourcing, dual-sourcing, or cutting capex would gut the growth segment, and Jabil has no pricing power over these buyers. (2) AI-capex cyclicality: Jabil has become a high-beta proxy for hyperscaler capex; the entire EMS group's re-rate (from ~12x to ~30x) is narrative-driven, and an "AI digestion" quarter would compress both EPS growth and the multiple simultaneously. (3) Margin is permanently thin: even after the mix shift, core OM is ~5.8% — Jabil keeps <6 cents of every revenue dollar; there is little buffer if utilization on the new $500M plant disappoints or component costs spike. Pre-mortem (18 months out, thesis broke): hyperscaler capex growth halved in 2H-FY27, Intelligent Infrastructure went flat, the new SE-US plant ran under-utilized (fixed-cost drag), the AP/consignment working-capital tailwind reversed (CFO fell below earnings), and the stock de-rated from ~29x to ~16x on a flat-to-down EPS — a ~45% drawdown despite "no company-specific scandal." That is the realistic bear path, and it requires nothing to go wrong at Jabil itself — only at its customers.
Is the multiple too high? At ~29x forward core for a cyclical low-margin manufacturer, yes on an absolute/historical basis (this is 2–3x JBL's own pre-2024 multiple); no relative to AI-EMS peers. The bet is entirely on whether the AI-infrastructure secular call holds long enough for EPS to grow into the multiple.
Lens 13 · Devil's Advocate (short-seller)
Dismantling the bull case:
- What structurally breaks the money machine: Jabil's incremental margin and growth both sit in Intelligent Infrastructure, which sits on a handful of hyperscalers' capex budgets. EMS is the most commoditized, lowest-margin, easiest-to-dual-source link in the AI chain — when hyperscalers want to squeeze cost, the contract manufacturer is the first place they look. Jabil's "moat" (Mikros/Hanley) is real but small relative to the $13.6B AI revenue base it's supposed to protect.
- Where revenue is concentrated and what happens if it shifts: 16% in one customer, ~half of revenue now in AI-DC. If that one customer moves volume to Foxconn (who has vastly more scale) or in-sources rack integration, JBL loses several points of revenue and the growth narrative in one stroke.
- Most dangerous competitor bulls underestimate: Hon Hai/Foxconn — an order of magnitude larger, already deep in AI servers (Nvidia partner), and capable of pricing Jabil out of hyperscaler rack business; and Celestica, which the market already crowns the purer AI-EMS play. Jabil is the third name in a two-name momentum trade.
- Worst capital-allocation critique: the buyback has hollowed out equity to $1.5B and is partly financed by a working-capital structure (consignment AP + AR securitization) that flatters cash flow. The EPS growth is real but amplified by financial engineering; strip the buyback and organic core-EPS growth is closer to mid-teens, not the headline.
- What must hold for $365: hyperscaler AI-capex must keep growing ~double-digits through FY27–28, the 16% customer must stay, the new plant must fill, and the ~29x multiple must not compress. Four things, all outside Jabil's control or at the mercy of one customer.
- If growth disappoints 20–30%: Intelligent Infrastructure +52% → +20% would still be "growth," but against this multiple and tone it would read as a miss; EPS growth halves, multiple compresses to low-20s,
30–40% downside. A genuine flat AI quarter → the mid-teens-multiple bear case ($200, ~45% down).
- Single scenario that permanently impairs: a structural hyperscaler shift to in-house rack integration (vertical integration away from EMS) — the same move Apple partly made — would permanently reset Jabil's addressable AI revenue. Plausibility: moderate over 3–5 years, low over 12 months.
Lens 14 · Management Questions (ordered by information value)
- Customer A is 16% of revenue in Intelligent Infrastructure — name the customer-concentration trajectory: is the top customer's share rising or falling, and what is the contractual protection (volume commitments, exclusivity, qualification lock-in) if they dual-source?
- What share of the ~$13.6B FY26 AI revenue is Mikros cooling + Hanley power (the higher-margin content) versus pass-through board/system assembly — and where does that mix go by FY28?
- Hyperscaler capex is the demand source: what specific signposts would tell you AI-DC demand is digesting, and how fast can you flex the new $500M SE-US plant's fixed costs if utilization disappoints?
- FY25 CFO was flattered ~$1.4B by accounts-payable/consignment timing and you run AR securitization — what is "clean" cash conversion excluding receivables-financing programs, and is it sustainable as revenue scales?
- You guide core OM >6% in FY27 — decompose it: how much is mix (Mikros/Hanley/Pii), how much operating leverage, how much new-plant ramp drag, and what utilization assumption underpins it?
- Capital allocation: with equity down to ~$1.5B and the stock at ~29x, why is buyback still the priority over building cooling/power capacity or larger value-chain M&A — at what price does buyback stop being the best use of FCF?
- What is the realistic ceiling on Intelligent Infrastructure as a share of total revenue before customer-concentration and cyclicality risk forces you to cap it?
- Foxconn and Celestica are scaling AI-server/rack manufacturing aggressively — where do you genuinely win against Foxconn's scale, and is it defensible on price?
- Networking & communications has lagged the cloud/DC growth — is that a share problem, a cycle problem, or a deliberate de-emphasis?
- Regulated Industries is the ballast — what core-margin and growth profile should we model for it through-cycle, and how much does Pii (pharma CDMO) move it?
- What is the through-cycle incremental margin on AI-DC revenue, and how should we think about decremental margins if that revenue contracts?
- Tariffs/trade policy: 75% of revenue is foreign-sourced with major China/Mexico exposure — what is the cost and timeline to reshore AI-DC production to the US, and who bears it?
- Where on the value chain do you draw the line — will you move further into power/thermal/IP (more margin, more capex) or stay an integrator?
- What is the succession depth below Dastoor now that Mondello has fully exited the board — is the bench built for a $40B+ AI-infrastructure company?
- If you could only protect one thing about this business over the next five years — the 16% customer, the cooling/power IP, or the buyback capacity — which, and why?