AT&S

Phase A — Understand the business

Lens 1 · Company Overview

AT&S makes the flat, layered interconnect that sits directly under and around a chip — high-end printed circuit boards (PCBs) and, increasingly, IC substrates (the FC-BGA / ABF package substrate that carries a processor die and fans its I/O out to the motherboard). It is not a chipmaker and not a foundry; it is the packaging-adjacent substrate supplier one layer below TSMC/AMD/Intel. Think of it as the "real estate the silicon is bolted onto."

Two reporting segments:

• Electronics Solutions (ES) — high-end PCBs (mobile, automotive, medical, industrial, aerospace) plus module/embedding technology. ~57% of external revenue in H1 FY25/26 (down from 61.4% a year earlier). Mature, cyclical, price-pressured; historically anchored by Apple (iPhone mainboards) and automotive. This is the cash cow.
• Microelectronics (ME) — IC substrates for CPUs, GPUs, AI accelerators and HPC, produced at Chongqing (China), Leoben (Austria) and now Kulim (Malaysia). ~43% of external revenue and rising fast — it grew 53% in constant currency in FY25/26 and 80% cc in Q4 alone at a ~36% segment margin. This is the growth engine and the entire reason the stock re-rated.

Business model & contract structure — the important recent shift: AT&S is moving from a build-it-and-hope PCB merchant to a "co-creation" model where customers pre-fund the capex. The Chongqing and Kulim expansions are "fully financed through long-term customer agreements" — customers pay AT&S to build the capacity they've committed to fill.. That is take-or-pay-like de-risking on the newest capacity, and it is the single most important structural change in the story. Whether it covers the whole €2B program or just the tranche is not disclosed.

Main customers: AMD (data-centre EPYC processor substrates at Kulim — the one named anchor), an unnamed "leading technology company" (industry sources: Intel, which gave AT&S its 2025 "EPIC Supplier" award), plus an undisclosed "key customer" driving the Chongqing expansion, and legacy Apple in PCBs. Main suppliers: Ajinomoto (ABF build-up film — ~95% global monopoly), plus glass-fabric (T-glass), copper foil, drill bits. Main competitors: Unimicron, Ibiden, Shinko Electric, Nan Ya PCB, Samsung Electro-Mechanics in substrates; Zhen Ding / TTM / Tripod in high-end PCBs.

Lens 2 · Supply Chain

Map, upstream → AT&S → end demand, with named stakeholders:

Upstream (inputs / chokepoints):

• Ajinomoto (Japan) — ABF (Ajinomoto Build-up Film), the dielectric layer that defines FC-BGA substrates. ~95% single-supplier monopoly. Ajinomoto informed substrate makers of a ~30% price hike effective Q3 2026 with the supply-demand gap extending to end-2027; lead times stretched to 20–30 weeks vs a normal 12–16. — the defining chokepoint of the whole industry, and it is upstream of AT&S, not controlled by it.
• Glass-fabric / T-glass (e.g. Nittobo-type high-end weave), copper foil, drill bits — all flagged as tightening. AT&S said on the Q4 call it hit T-glass shortages but "very successfully managed" them via supplier qualification and allocation.
• Capital equipment: laser drilling, plating, lithography tools (generic semi-equipment vendors; not disclosed).

AT&S (the node): converts film + laminate + copper into substrates/PCBs at Leoben & Fehring (AT), Chongqing (CN), Kulim (MY), Nanjangud (IN), Ansan (KR — SOLD Jan 2025).

Downstream (customers → end market):

• Substrates → AMD (EPYC/Instinct), Intel (unconfirmed second anchor), undisclosed Chongqing customer → assembled into AI servers / data-centre CPUs & GPUs → bought by hyperscalers (the true end-demand: Microsoft, Meta, Amazon, Google, xAI capex).
• PCBs → Apple (iPhone mainboards, ~30% of iPhone board volume per one estimate), automotive OEMs, medical/industrial.

Chokepoints & single-source dependencies: (1) ABF film from Ajinomoto is the binding upstream constraint — good for pricing power industry-wide, bad because AT&S can't secure more output than Ajinomoto allocates it. (2) AT&S is itself a near-single-source qualified supplier for specific AMD parts once designed in — high switching cost both ways. (3) End-demand is concentrated in AI data-centre capex, a single macro driver. Names present throughout — this lens is grounded.

Lens 3 · Competitive Advantages (moats)

What the moat actually is: AT&S is one of only five credible ABF/FC-BGA substrate makers on earth — Unimicron, Ibiden, AT&S, Nan Ya PCB, Shinko together hold ~74% of global ABF-substrate share, and AT&S is the only European one. Qualifying a new FC-BGA substrate line at a customer like AMD takes years; AT&S built and got Kulim AMD-certified in ~record time (2yr build + 1yr ramp). That combination — scarce qualified capacity + multi-year design-in lock-in + Western/diversified geographic footprint in a Taiwan/Japan-concentrated industry — is a real, if narrow, moat.

Durable sources:

• Switching costs / qualification lock-in (strong once designed into a socket; the substrate is co-developed with the chip).
• Scarcity of capacity in an oligopoly where ABF supply is the binding constraint through 2027.
• Geographic diversification — a genuine differentiator as AMD/Intel/hyperscalers actively de-risk away from Taiwan concentration; AT&S's Europe+Malaysia footprint is a supply-security selling point.
• Emerging tech optionality — glass-core substrates (announced for AI/HPC/photonics), embedding technology, 40kW-class high-power substrates.

Where the moat is weak (bull-beware): AT&S is the #5–ish player, not #1; Unimicron ($24–42B mkt cap depending on source) and Ibiden dwarf it in scale and balance-sheet firepower. Its PCB heritage segment faces "persistent high price pressure." The customer holds real power — AMD's design-in made the Kulim ramp, and AMD's book could unmake it. Bargaining power over Ajinomoto is essentially nil (95% monopoly upstream). This is a scarce-but-subordinate position in the value chain, not a dominant one.

Lens 4 · Segments

Revenue by segment (external / third-party). No segments.csv on the shelf — all ``.

Trend and cause — this is the whole thesis in one table. ES (the mature PCB base) is flat-to-soft — smartphone seasonality, Chinese-New-Year effects, price pressure. ME (substrates) is accelerating hard — pure AI-infrastructure pull, and it carries a ~36% margin vs the group's 23.3%, so every point of mix shift toward ME lifts group profitability. The mix flip from ~39% → 43%+ ME in one year, at ME's fat margin, is exactly what drove the FY25/26 EBITDA margin up +560bps to 23.3%. By geography, ME is Malaysia (Kulim, AMD) + China (Chongqing) + Austria (Leoben); ES is Austria + China + India, Apple/auto-weighted. Accelerating, and the acceleration is the good kind (high-margin AI). The risk is that the flat 57% (ES) is where the cyclical downside lives if the consumer/auto cycle rolls.

---

Phase B — Measure performance

Lens 5 · Earnings Result — FY2025/26 (year ended 31 Mar 2026, reported May 2026)

The turnaround print.

• Revenue €1.8bn (reported €1,791m), +21% constant-currency (FX headwind ~€45m).
• EBITDA €418m, margin 23.3%, +560bps YoY, a +49% jump.
• EBIT €66m, +238% YoY — still thin because depreciation from the Kulim/substrate build is now hitting the P&L.
• Net result: still a small loss. One third-party table shows net income −€25.6m, basic EPS −€1.11 for FY ending Mar-2026. The gap between +€418m EBITDA and a negative bottom line is depreciation (€330m→€400m) + interest (€80m/yr) + hybrid coupon — i.e. the ramp is EBITDA-positive but not yet clean-net-income-positive. This matters: the headline "record EBITDA" coexists with a net loss.
• Operating free cash flow +€236m — a ~€700m swing from −€498m the prior year. The genuinely bullish number: capex rolled off the FY22–24 mega-build and the substrate plants started paying.
• Balance sheet: net debt/EBITDA 3.2x (from 5.3x), equity ratio 22.6%, >€900m cash + unused lines. Still levered, but de-levering fast.
• Market reaction: the stock did not just pop on the print — it ~10x'd over the year (1-yr change +909.75% ) and hit an all-time high €210 on 15 Jun 2026 on the follow-on capacity/guidance raise. What was priced in: nothing short of a full AI-substrate breakout.

Unusual vs its own history: AT&S swung from guidance-suspended and −€498m FCF (FY24/25 trough) to record EBITDA + positive FCF (FY25/26) in a single year. That is a violent cyclical+structural inflection — exactly the kind that gets extrapolated too far.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the shelf; sentiment reconstructed from Q1→Q4 FY25/26 call coverage.

• Q1 FY25/26 → Q3 FY25/26: tone shifts from "recovering / AI focus" to "revenue surge, AI focus" — visibly warming as the ME ramp beats.
• Q4 FY25/26 (May 2026): euphoric. New CEO Michael Mertin calls it "just the beginning of a very interesting journey," frames AT&S as moving from "traditional PCB provider to innovation partner," and repeatedly notes demand exceeding supply ("many places where we can't fulfil… customers are wishing" for more). CFO Gerrit Steen elevates operating free cash flow to a "key KPI."
• June 2026 (guidance raise): management lifts FY26/27 guidance twice within weeks — the loudest possible confidence signal.

What they now emphasise (new vocabulary): "co-creation," "customer-financed expansion," "embedding technology," "glass-core," "40 kilowatt high-power substrates," "innovation partner." What dropped out: the FY23–24 language of "adjusting growth pace to market conditions" and "postponing medium-term targets." Sentiment trend: sharply, almost promotionally, positive — a full 180° from the 2023–24 apologetic tone. The flag: this is a management team that was walking targets back two years ago and is now raising them twice a quarter into a doubling stock. Enthusiasm is warranted by the numbers; the degree of it is a thing to watch (Lens 13).

Lens 7 · Comps

Peer set = the ABF/FC-BGA substrate oligopoly (the right peers — not the foundry/GPU names the hardware index lists, which are a different business). Multiples `` or n/a.

AT&S valuation math ``: at ~€200/sh × ~38.8m shares ≈ €7.76bn equity + ~€1.34bn net debt (3.2× × €418m EBITDA) ≈ ~€9.1bn EV. That is ~22x trailing EV/EBITDA (on €418m) and ~5x trailing EV/Sales — falling to ~10x EV/EBITDA and ~3.4x EV/Sales forward only if you grant the raised FY26/27 guidance (rev €2.6–2.8bn, EBITDA ~€0.9bn).

Read: the substrate group genuinely trades rich (Ibiden 34x, Shinko 14x EV/EBITDA) on the AI supercycle, so AT&S's ~22x trailing isn't an outlier for the sector. But AT&S is the sector's smallest, most-levered, net-loss-making member being valued above where Shinko (profitable, mature) trades and near the top of the pack — and, decisively, the sell-side consensus target is ~€74.67 (range €35–133) versus a ~€200 tape. The market is ~2.7x above its own analysts' fair value. No fabricated multiples — where a peer figure wasn't sourced it's marked n/a.

Lens 8 · Stock-Price Catalysts (moves >5%, last ~5 years)

Pattern of what actually moves ATS.VI:

• FY22 (cal-2021→22): revenue ramped to a then-record (~$2.02bn cal-2022) on the first substrate build-out — stock strong.
• Mar 2023: "adjusts growth pace… postpones medium-term targets by one year" — a de-facto profit warning; multiple compression.
• FY24 cyclical trough (2023–24): semi/smartphone downturn, substrate start-up costs, guidance suspended in early FY25/26 on "volatile order behaviour of a key customer," −€498m FCF, net-debt/EBITDA spiking to ~6.1x (Dec-24) — stock in the doldrums (~€17–20 mid-2025).
• Sep 2024–Jan 2025: Korea (Ansan) plant sale to SOMACIS, ~€405m (+€17m interest; ~€386m net cash) — deleveraging catalyst, took net-debt/EBITDA to ~2.5x.
• May 2025: Kulim high-volume manufacturing starts for AMD — the de-risking event that flipped the narrative. New CEO Mertin starts.
• 2025→2026: the ~10x melt-up. ~€19 (Jul-25) → ~€100 (Apr-26) → ATH €210 (15 Jun 2026) on the €2B AMD/Intel expansion + double guidance raise. 1-yr +909%.
• 30 Jun 2026: founding foundations sell 567,000 shares at €200 (see Lens 9) — a top-tick insider distribution.

What the market reacts to: (1) substrate/AI proof-points (Kulim ramp, customer deals) — massive upside reactions; (2) guidance credibility (2023 reset = down hard; 2026 raises = up hard); (3) deleveraging events (Korea sale). The tape has become a high-beta AI-substrate call option — it now moves on AI-capex sentiment far more than on its own steady PCB base.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO — Dr. Michael Mertin (since 1 May 2025, 3-yr term). Physicist (RWTH Aachen; PhD laser processing, Fraunhofer ILT), 10 years at Carl Zeiss, then CEO of JENOPTIK AG 2007–2017 (a respected German photonics/optics scale-up). Track record: real operator, industrial-tech pedigree, not a promoter by history — but <15 months in the seat, so the euphoric FY25/26 narrative is largely inherited momentum he is now amplifying.
• CFO — Gerrit Steen (since 1 Feb 2026 — brand new). Ex-Group CFO of DAMAC (Dubai real estate/tech), earlier Fresenius Kabi (Global CFO / Asia CEO) and Heraeus. Elevating operating-FCF to "key KPI" is the right instinct for a post-mega-capex balance sheet. Both top executives are <18 months tenured — the leadership running the raised guidance is new to this company.
• Ownership / skin in the game: controlled by two founding-family foundations — Androsch Privatstiftung ~15.4% and Dörflinger Privatstiftung ~12.1% (+ Dörflinger Management ~5.9%), ~36% combined; free float ~48% retail-heavy. High insider ownership is normally a positive alignment signal. BUT — critical, dated 30 Jun 2026 — the two foundations sold 567,000 shares at €200/share, trimming the combined stake ~1.5pts (from ~36%), reported via "market insiders," no stated rationale.. Founders selling into the all-time high, days after their own company's second guidance raise, is the loudest capital-allocation signal in this dossier — see Lens 13.
• Capital-allocation history — genuinely mixed. The FY22–24 substrate mega-build (~€1.8–1.9bn combined capex; €996m in FY22/23 alone) blew out leverage to ~6x and forced the Korea disposal and a suspended dividend — classic over-build into a cycle top. The offset: it created the exact asset (AMD-qualified Kulim) that is now printing — so the aggressive bet arguably paid off, but only after a near-covenant scare. The new "customer-financed capex" model is a real discipline improvement.
• Red flags: (1) The foundation share sale at the top. (2) Governance history: in a prior era Androsch forced out CEO Gerstenmayer after a failed capital increase — a foundation willing to defenestrate a CEO in a power struggle. (3) Serial target-resetting (2023 postponement; suspended FY25/26 guidance) — a credibility ledger that argues for discounting the current bullish guide. (4) Two brand-new top executives owning an extrapolated narrative.
• Archetype: foundation-controlled European industrial, now run by professional managers (Mertin/Steen), not founder-operators. Implication: competent, disciplined, but the owners (foundations) are behaving like sellers, not compounders, right now.

Lens 10 · Forensic Red Flags

Forensic lens. All / — no filings on the shelf to tie to line items, which is itself a limitation to state plainly: I could not audit the cash-flow statement directly.

• Earnings vs cash quality: the good news is cash is better than earnings here (OpFCF +€236m vs a small net loss) — the opposite of the usual red flag. The €418m EBITDA → negative net income gap is depreciation + interest + hybrid coupon, not aggressive accruals. Watch that EBITDA is doing heavy lifting in a company that is still GAAP-loss-making; "record EBITDA margin" headlines can mask that shareholders earned nothing net in FY25/26.
• Depreciation & capitalisation: €1.8–1.9bn of recent capex is now depreciating (€330m→€400m/yr). Capitalised-cost/useful-life assumptions on brand-new substrate lines are a judgement area — if utilisation disappoints, impairment risk on Kulim Plant 2 (shell-ready, un-equipped = potential sunk cost).
• Leverage / hybrid: a €350m hybrid bond (Jan-2022, 5.0% coupon, first call 2027) is treated as equity but resets/steps up in early 2027; management flagged a refinancing hybrid issue for it. Equity-credit treatment flatters the equity ratio (22.6%).. Not fraud — but a real 2027 refinancing event on a still-sub-30% equity ratio.
• Segment reporting: only two segments, and the market cares almost entirely about ME — reasonable, but ME margins (~36%) are new and ramp-dependent; a couple of quarters is a thin track record to underwrite the re-rating on.
• Receivables/inventory vs revenue, related parties, contingencies: not verifiable from the shelf (no 10-K/annual-report line items ingested) — flag as an open audit item; the Austrian annual report (ats-report.net) would be the primary source.

Regulatory findings (required sub-section). Read from regulatory/regulatory-findings.md (Step 0 output):

• SEC (EDGAR LR + AAER): none possible. AT&S has no CIK and does not file with the SEC; the regulatory-findings file confirms 0 SEC findings and notes no EDGAR search is applicable.
• Non-SEC enforcement (web): the guided search ("AT&S" (FTC OR DOJ OR FDA OR CFPB OR "consent decree" OR settlement OR fine OR penalty)) surfaced no material enforcement actions, fines, or consent decrees against AT&S in the current pass.
• 10-K Item 3 (Legal Proceedings): not available — no US-style 10-K exists for this filer; the equivalent disclosure lives in the Austrian annual report, not on the shelf.
• Verdict: No material regulatory or legal findings for AT&S — verified via SEC EDGAR EFTS (LR/AAER, 0 hits, no CIK) and web search as of 2026-07-06. The governance concerns (foundation share sale, historic CEO ouster) are corporate-governance signals, not regulatory findings.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (EPS, next 3 fiscal years)

Built bottom-up from FY25/26 actuals + management's raised FY26/27 guidance. All ``; no forecast.ts create logged (unattended --watchlist run per skill rules). ~38.8m shares. EUR.

Anchors: FY25/26 rev €1.79bn, EBITDA €418m (23.3%), EBIT €66m, net ≈ −€26m, D&A ≈ €350m, interest ≈ €80m. Raised FY26/27 guide: +45–55% cc revenue, 32–37% EBITDA margin, €1.0–1.2bn capex.

The base-case scoreable call (not logged, per rules): "ATS.VI FY26/27 (Mar-27) net-GAAP EPS ≥ €6.0" — my `` p ≈ 0.45 (guidance is aggressive and this management's guide-history is poor; but the AMD/Ajinomoto demand backdrop is real). The number that actually matters isn't EPS — it's utilisation of Kulim Plant 2 and whether the €2B capex stays customer-financed. Runway is fine (>€900m cash + customer-funded capex + 2027 hybrid refinanced), so this is a return-on-capital question, not a survival question.

Lens 12 · Bull vs Bear

Bull case. AT&S is a scarce, Western, AMD-qualified ABF-substrate maker at the exact moment the AI packaging stack is exploding (layer counts 3+3 → 11+11 → 13+13; ABF in structural shortage to end-2027; Ajinomoto raising prices 30%). It has flipped from cash-burning over-builder to +€236m FCF with a 23.3% EBITDA margin, its high-margin (~36%) substrate segment is compounding 50%+, and — crucially — its next €2B of capacity is customer-financed, so growth no longer means balance-sheet risk. Management raised guidance twice in weeks (to +45–55% rev / 32–37% EBITDA). If Kulim P2 fills and glass-core lands, this is a €4bn-revenue, €1bn-EBITDA company inside three years, and the only non-Asian play on a supply-security theme AMD/Intel/hyperscalers are actively paying for. Contrarian upside the market may still under-price: the customer-financed capex model structurally de-risks the historic knock on AT&S (it over-builds and blows up leverage) — this time the customers carry the risk.

Bear case (permanent-impairment risks). (1) Valuation, not business, is the problem. At ~€200 the stock is ~22x trailing EV/EBITDA, net-loss-making, and ~2.7x above the €74.67 sell-side consensus. It is priced for the bull path to be delivered, not attempted — by a management team that has missed and reset guidance repeatedly and is <18 months tenured. (2) AI-capex cyclicality / air-pocket: substrate demand is a derivative of hyperscaler capex — a single digestion quarter would de-rate a 10x'd stock brutally, and Kulim Plant 2 is shell-ready but un-equipped (sunk cost if unfilled). (3) Upstream squeeze: Ajinomoto's 30% ABF hike + T-glass/copper shortages hit COGS with a "couple of months" pass-through lag — margin can compress before price catches up. Pre-mortem (18 months out, thesis broke): AI order rates normalised in H2-2026, AMD pushed out its second-source ramp, Kulim P2 sat idle and took an impairment, the ABF cost squeeze bit before pricing caught up, the 2027 hybrid refinanced expensively — and a stock that had priced flawless execution round-tripped from €210 toward the €75 consensus. The founding foundations, who sold at €200, look prescient.

Are multiples too high? For the business, no — a scarce AI-substrate asset deserves a premium. For the stock at this price, yes — it discounts the top of guidance from a serial guidance-misser, with owners distributing stock into it.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks it: AT&S sells a derivative of one macro variable — AI data-centre capex — through a handful of customers (AMD + one unnamed). Revenue concentration is severe: lose or see a push-out from AMD and the ME growth engine stalls with a half-built Kulim P2 depreciating against it. The moat (qualification lock-in) cuts both ways — it's also a single-customer dependency on the parts it's designed into.
• Most dangerous competitor bulls underestimate: Ibiden — vastly larger, premium-margin, and it announced (Feb-2026) a big AI-substrate capacity add plus a $1.2bn Arizona flip-chip fab for late-2027. When the #1/#2 flood qualified capacity in 2027–28, AT&S's scarcity rent compresses and it's back to being the sub-scale #5. Unimicron (#1) similarly can out-invest it.
• Worst capital-allocation / governance tells: the FY22–24 over-build to ~6x leverage (needed a plant sale to survive), a foundation that ousted a prior CEO in a power struggle, and — the tell that should stop any bull cold — the controlling foundations selling 567k shares at €200 into their own guidance raise, with no stated reason. Insiders who know the order book best are taking chips off the table at the top.
• What must hold for €200 to be right: +45–55% revenue and 32–37% EBITDA and Kulim P2 fully utilised and ABF cost pass-through and no AI-capex digestion and clean hybrid refinancing — six things, all at once, from a team that has serially guided down.
• If growth disappoints 20–30%: FY26/27 revenue lands ~€2.1–2.25bn (bear column), net EPS collapses toward €1, and a stock at ~€200 on a net loss has no earnings floor — the €74.67 consensus and even the €35 low become the gravity. A 50–65% drawdown is entirely available without the business failing — just on de-rating from priced-for-perfection to priced-for-reality.
• The single scenario that permanently impairs: a 2027–28 substrate-capacity glut (Ibiden AZ + Unimicron + everyone's AI adds land together) coinciding with an AI-capex pause — turning today's scarcity rent into idle depreciation across the sector. Plausibility: medium for the de-rate, lower for permanent impairment (the AMD relationship and Western footprint have lasting value). Short thesis is about the price, not the company.

Lens 14 · Management Questions (ordered by information value)

1. Of the €2.0bn Kulim + Chongqing capex program, exactly what fraction is contractually customer-financed (take-or-pay or prepayment), and what are the volume/duration commitments — for Kulim Plant 2 specifically?
2. Beyond AMD, name the revenue concentration: what % of Microelectronics FY25/26 revenue was the top customer, and top two? What is AMD's share?
3. Is the unnamed "leading technology company" Intel, and what stage is it — qualified/ramping/design-in only?
4. Utilisation of Kulim Plant 1 vs Plant 2 today, and the utilisation you need for the 32–37% FY26/27 EBITDA margin to hold?
5. Given Ajinomoto's 30% ABF price increase and the "couple of months" pass-through lag, what is the FY26/27 gross-margin drag before pricing catches up, and are ABF volumes contractually allocated to you?
6. Why did the Androsch and Dörflinger foundations sell 567,000 shares at €200 on 30 June — and have they signalled further sales or a lock-up?
7. FY26/27 guidance implies EBITDA ~€0.85–1.0bn; at what revenue/utilisation does the group turn clean-GAAP-net-income-positive, and in which quarter?
8. The €350m hybrid resets/first-calls in early 2027 — refinancing coupon assumption, and pro-forma equity ratio after the new hybrid?
9. When do Ibiden's Arizona fab and the industry's other AI-substrate adds bring competing qualified capacity online, and what does that do to your pricing/share in 2027–28?
10. What is maintenance (vs growth) capex for the substrate business at steady state — 5% or the 8–10% of sales some analysts model — and thus normalised FCF conversion?
11. Glass-core substrates: revenue timeline, which customers, and is it incremental or a cannibalising transition from ABF?
12. How exposed is the Electronics Solutions (PCB) base to Apple specifically, and what is the plan as that segment stays flat-to-declining?
13. Chongqing (China) geopolitical exposure — what share of substrate capacity sits in China, and how do you hedge US-export / tariff risk on AI parts?
14. With two executives <18 months tenured, what changes vs the prior regime's capital-allocation and guidance philosophy — and why should the market trust this raised guide after the 2023 reset?
15. What would have to go wrong for FY26/27 to miss the low end (+45%) of the raised guidance — the top two risks you actually worry about internally?

---