IQE plc

Phase A — Understand the business

Company Overview

IQE plc is the world's leading outsourced (merchant) supplier of compound-semiconductor epitaxial wafers — "epiwafers." It does one thing and does it at the front of the value chain: it grows atomically-precise crystalline layers of III-V materials (GaAs, InP, GaN, GaSb) onto substrates using MOCVD (metal-organic chemical vapour deposition) and MBE (molecular beam epitaxy), then ships those epiwafers to chipmakers who fabricate the actual devices. It does not make finished chips — it is the layer below the foundry. Think of it as the "epitaxy foundry" for everyone who can't justify owning their own reactors.

What it sells, by end-market:

• Wireless (RF): GaAs and GaN epiwafers for power amplifiers in mobile handsets, 5G infrastructure, and defence radar. FY2025 revenue £40.1m.
• Photonics: GaAs-based VCSELs (the lasers behind smartphone 3D-sensing / Face ID), and increasingly InP epiwafers for the lasers and modulators inside AI-datacentre optical transceivers (800G / 1.6T), plus infrared imaging and military sensing. FY2025 revenue £57.1m.
• CMOS++: legacy advanced-silicon programme — effectively dead: revenue collapsed from £863k (2024) to £65k (2025) after IQE closed its South Wales silicon site.

Differentiation: IQE describes itself as the only pure-play epiwafer producer with the full III-V material portfolio AND global manufacturing across all three regions (US, EU-UK, Asia). That tri-regional footprint is a real and rare asset in a world re-shoring its chip supply chain — but it is also the source of IQE's cost problem (three sub-scale fabs instead of one big one).

Customers / contract structure: Historically concentrated and "you eat what the smartphone cycle gives you." Anchor relationships: Lumentum (long-term VCSEL/3D-sensing supply, extended) and now MACOM (April 2026 long-term supply agreements for InP + GaN, alongside MACOM taking equity) and Tower Semiconductor (June 2026 multi-year InP supply agreement). Contracts are increasingly minimum-volume commitments rather than pure spot — a structural improvement (see Lens 3). End-buyers historically trace to Apple (via the VCSEL chain — IQE is a long-rumoured/known Apple 3D-sensing supplier) and the major optical-transceiver OEMs (Coherent/Lumentum, Broadcom-adjacent).

Headline shape: A structurally important, technically excellent, chronically unprofitable small-cap (FY2025 revenue ~£97m; adjusted EBITDA just £3.2m) that in 2025 went into a formal strategic review / sale process and in 2026 was rescued by an £81m financing anchored by a strategic customer. The business is real; the equity has been a value-trap for a decade.

Supply Chain

IQE sits near the top of the AI / mobile-optics supply chain — upstream of the foundries, downstream only of raw substrate and tool makers. Mapping it with named stakeholders:

Upstream (inputs into IQE):

• Substrate suppliers — IQE buys bare wafers (GaAs, InP, GaN-on-Si, sapphire/SiC) and grows epi on them. The critical chokepoint is InP substrate, which is genuinely scarce: AXT Inc. and Sumitomo Electric are the dominant InP/GaAs substrate makers. IntelliEPI's CEO publicly called InP shortage "a bottleneck for the entire AI infrastructure" in Q1 2026. IQE's access to InP substrate is a gating input for its single best growth story.
• Tool/reactor makers — Aixtron (MOCVD), Veeco (MOCVD/MBE), and specialist MBE houses supply the reactors. Capex-heavy and long-lead.
• Precursor gases / metalorganics — trimethylgallium, arsine, phosphine; standard semiconductor chemical supply.

IQE (the transformation step): epitaxy across MOCVD + MBE, three regions — Cardiff (UK, HQ + largest site), Massachusetts (US, ex-MBE Technologies / Greensboro NC area operations), and Taiwan (the operations IQE has been trying to sell)..

Downstream (IQE → device → system):

• Device fabs / foundries: WIN Semiconductors (the ~65%-share GaAs RF foundry — both a customer-adjacent processor and the gorilla of the layer below), Tower Semiconductor (silicon-photonics fab — now an IQE InP customer via the June 2026 deal), MACOM (buys InP/GaN epi, processes into RF + optical components).
• Component OEMs: Lumentum, Coherent, Broadcom (optical + 3D-sensing), turning epi → lasers/PAs/transceivers.
• System buyers / end demand: Apple (3D sensing in iPhone — the swing factor in IQE's photonics history), the hyperscalers (Nvidia/AMD AI clusters needing 800G/1.6T optics), defence primes (US mil/defence infrared + radar), handset OEMs (Samsung, Chinese Android).

Chokepoints & single-source dependencies:

1. InP substrate scarcity (upstream) — the binding constraint on IQE's AI-optical upside. IQE's value rises because InP is scarce, but IQE itself is exposed to that same scarcity on the buy side.
2. Customer concentration (downstream) — historically Apple-chain photonics + a handful of RF customers. The 2026 MACOM/Tower deals deepen concentration even as they de-risk it commercially.
3. Reactor capacity (internal) — epitaxy capacity is the asset. Under-investment (FY2025 capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. only £5.1m) means IQE can be supply-constrained exactly when AI-optical demand inflects — a self-inflicted chokepoint born of a starved balance sheet.

This lens is the whole thesis: IQE's strategic value is being a tri-regional, full-portfolio epi node in a re-shoring, InP-starved AI-optics chain. Its strategic weakness is that it has been too poor to buy the reactors that value requires.

Competitive Advantages (moats)

The moat is real but narrow, and it has been eroding at the edges:

• Process / know-how moat (strong). Epitaxy is a black art — yield, uniformity, and defect density on III-V crystal growth are hard-won and embodied in tacit process knowledge and trained engineers. IQE has decades of it across the full III-V portfolio. This is the durable core. New entrants can't buy a reactor and match IQE's VCSEL or InP yields in a year.
• Portfolio breadth + tri-regional footprint (genuinely differentiated). IQE is the only merchant player offering GaAs + InP + GaN + GaSb across MOCVD and MBE, in US + Europe + Asia. In a re-shoring world ("I need a non-China, non-single-region epi source"), that footprint is a strategic asset MACOM literally paid equity to lock up.
• IP estate (now cleaner). IQE holds porous-silicon patents — and in June 2026 settled its 2022 lawsuit against Tower Semiconductor, granting IQE a broad worldwide royalty-free license, removing a legal overhang.
• Switching costs (moderate). Qualifying an epiwafer supplier into a device fab's process is slow and expensive; once IQE is designed-in (Lumentum VCSELs, MACOM InP), it is sticky. The new minimum-volume long-term supply agreements convert that stickiness into contracted revenue — a structural upgrade from the old spot-exposed model.

Bargaining power — who needs whom? Historically IQE was the weaker party: a sub-scale supplier to giant customers (Apple-chain) who could dual-source or in-source. The 2026 reversal is subtle and important — MACOM and Tower needed to secure non-Asian InP epi capacity badly enough to invest/contract, which shifts power toward IQE for InP specifically. But on GaAs RF wireless, IQE remains the weak party against WIN Semiconductors (~65% foundry share) and a commoditising handset market.

Where the moat is weakest: Wireless/GaAs is increasingly commodity, Asian-clustered (Asia-Pacific = ~60% of the GaAs wafer market ), and price-competitive. IQE's edge there is shrinking; its edge in InP photonics + GaN power + defence is where the durable moat lives.

Segments

No our figures in the research layer (web-only). Reconstructed from disclosures:

The story the segments tell: IQE is mid-pivot from a cyclical RF/3D-sensing wafer maker into an AI-optical InP play. Wireless (the old core, ~57% of 2024 revenue) is shrinking and commoditising; Photonics (now the larger segment at 59% of FY2025) is the growth vector and the reason for the MACOM/Tower deals. The 2026 guidance (>20% revenue growth ) is a bet that Photonics+defence acceleration outruns any further Wireless erosion.

Geography: Not cleanly disclosed in segment form; manufacturing is UK (largest) / US / Taiwan, with end-demand skewed US (defence + hyperscale) and Asia (handset). The pending Taiwan operations sale would shrink the Asian footprint and the revenue base — a deliberate "shrink-to-core" move.

---

Phase B — Measure performance

Earnings Result (FY2025, reported 28 May 2026)

The defining print of the recovery narrative — weak on the surface, inflecting underneath [all figures web: semiconductor-today 2026-05-29; Investegate/RNS FY2025; AJ Bell]:

• Revenue £97.3m, −17.6% YoY (FY2024 £118.0m). Landed in the middle of the £90–100m guided range — no further negative surprise, which for IQE counts as a win.
• Adjusted EBITDA £3.2m (3% margin) vs FY2024 £8.1m (7% margin) — margin halved on lower volume / negative operating leverage. EBITDA still positive, which kept the going-concern wolf from the door.
• Operating cash flow £8.1m, up sharply from £1.3m — working-capital release (inventory normalising) flattered cash even as P&L weakened. A genuine positive.
• Cash £15.7m (up from £4.7m); adjusted net debt £31.5m (up from £18.8m) — net debt rose despite higher cash, because gross borrowing (RCF + March-2025 convertibles) grew. This is the balance-sheet stress that forced the strategic review.
• Capex just £5.1m — starvation-level for an epitaxy business; mostly GaN/power + micro-LED seed investment. Under-investment is the chronic disease.
• H1 2025 was ugly: revenue £45.3m (H1 2024 £66.0m), pretax loss £26.5m (vs £13.9m prior), Wireless −52%. H2 then beat (Jan 2026 trading update flagged strong defence/AI/datacentre/handset demand) — the H1-disaster / H2-recovery shape is the key tell.

Guidance / outlook: FY2026 revenue to exceed +20% YoY (→ ~£117m+ ); adjusted EBITDA targeted high-single-digit to low-double-digit £m. Tone shifted decisively positive vs the going-concern language of mid-2025.

Market reaction: The stock has been violently re-rating on the recovery + deal flow — +22.6% on 15 June 2026 (the Tower deal day) to ~56.7p. Over 2025 it had been a wipeout (down ~85–89% at the October-2025 lows per MarketBeat alerts), so this is a dead-cat-or-real-turn debate playing out in real time.

Unusual vs own history: Positive operating cash flow with a deepening pretax loss (working-capital driven), and net debt rising while cash rises (gross-debt driven) — both flag a company being run for survival/liquidity, not earnings.

Earnings Calls (sentiment trend)

No transcripts in the research layer; reconstructing management tone from disclosures across the cycle:

• 2023 (inventory glut): defensive, damage-control — "de-stocking," "acceleration of de-stocking trends," guidance cuts, headcount −10%, site closures (Singapore 2022, Pennsylvania 2023). Tone: survival.
• H1 2024: cautious-optimistic — "inventory normalising," "Q1 in line," recovery hoped-for. Tone: stabilising.
• Late 2024: regime break — CEO Americo Lemos departs with immediate effect (Oct 2024), CFO Jutta Meier steps in as interim, stock plunges. Tone: crisis / leadership vacuum.
• Mid-2025: strategic review widened to a potential sale of the whole company; Taiwan sale negotiations; "additional early-stage expressions of interest." Tone: for-sale, distressed.
• 2026: transformation language — "transformational £81m," "tier-1 hyperscale and AI infrastructure," "scalable high-volume manufacturing," Jutta Meier confirmed permanent CEO, MACOM/Tower deals. Tone: re-platformed, growth.

What they stopped saying: "de-stocking," "inventory glut," "strategic review / sale of the company." What they started saying: "AI infrastructure," "hyperscale," "InP," "long-term supply agreements," "strategic partner." The phrase-shift is the cleanest evidence the narrative has genuinely turned — though narrative turns at IQE have proven false before (the 2024 "recovery" reversed into the 2025 collapse).

Comps

Read: At ~0.9x sales IQE looks optically cheap for an "AI-optical" name — but that multiple is correct for a sub-scale, loss-making, capex-starved foundry with a decade of value destruction. A re-rate requires the earnings to show up, not just the revenue. Note: I could not source clean EV/Sales, EV/EBIT, P/E, dividend yield, or 5-yr ROE for the peer set — IQE pays no dividend, has negative/erratic ROE, and the peers are themselves mostly unprofitable. Writing n/a rather than fabricating. The honest comp statement: there is no good comp; IQE is a special-situation recovery option, not a multiple-relative trade.

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

IQE is a high-beta, headline-whipped AIM small-cap. The pattern of what moves it:

• 2021: VCSEL/3D-sensing demand + supply-chain tightness → multi-year highs (~50p+).
• Nov 2021 / 2022: Apple profit-warning read-through ("VCSEL suppliers hit by Apple profit warning" ) + the inventory cycle rolling over → sustained de-rate.
• 2022–2023: inventory-glut profit warnings, site closures, guidance cuts → −66% in 2023 alone (50.2p → ~17p).
• Oct 2024: CEO Lemos abrupt exit → sharp plunge.
• 2025: strategic review → potential sale of company + going-concern stress + dilutive March-2025 convertibles → down to ~6–7p, −85% to −89% at Oct-2025 lows.
• Apr–May 2026: MACOM £45m strategic investment + £81m total raise → violent recovery bounce.
• 15 Jun 2026: Tower Semiconductor InP supply deal + IP settlement → +22.6% in a day.

What the tape reveals: The market reacts to (1) Apple/handset demand signals, (2) the inventory cycle, (3) balance-sheet / dilution / going-concern events, and now (4) AI-optical deal flow (MACOM, Tower). It is not an earnings-multiple stock — it is a demand-cycle + balance-sheet + strategic-event stock. Caveat: the 2025 share consolidation/raise re-based the price; pre-2026 pence levels are not directly comparable to today's ~47–57p — see Lens 7 conflict note.

---

Phase C — Judge people & books

Management

• CEO: Jutta Meier — joined IQE Jan 2024 from Intel as CFO; became interim CEO Oct 2024 on Lemos's abrupt exit; confirmed permanent CEO May 2025, holding the dual CEO+CFO role during the review. Track record: too early to judge as CEO — but she navigated the company through a going-concern crisis to an £81m rescue and two strategic-customer anchor deals (MACOM, Tower) inside ~18 months. That is a credible distressed-turnaround execution even if the destination is uncertain. Intel pedigree = operational/financial discipline, not visionary growth. The CEO+CFO dual-hat is a red flag for governance (concentration of control during a crisis), defensible only as a transitional posture.
• Executive Chair: Mark Cubitt — joined as Chair-elect, became Executive Chair to backstop Meier through the review; Phil Smith (chair since 2019) stepped back to NED. An executive chair alongside a dual-role CEO/CFO is a heavy concentration of authority — appropriate for a crisis, must unwind for the equity to be institutionally ownable.
• New board (2026): MACOM representatives Robert Dennehy and David O'Carroll join post-investment. This is the single most important governance fact: IQE's largest strategic shareholder (~11.5%) and a major customer now sits on the board. Aligns IQE's roadmap with MACOM's InP/GaN demand — good for execution focus, but it subordinates minority shareholders' interests to a strategic's and effectively makes MACOM a kingmaker in any future sale.
• Capital-allocation history (the indictment): A decade of value destruction. Repeated dilutive raises, three sub-scale fabs never consolidated, chronic under-investment in the one asset that matters (reactors), site closures (Singapore, Pennsylvania, South Wales) that read as retreat not strategy. ROE/ROIC structurally negative-to-poor. The prior regime over-promised on 3D-sensing/Apple and never converted scale into profit.
• Founder vs professional manager: Fully professional-manager (founder-era long gone). Current team is a restructuring/turnaround crew, not founders compounding a vision. Implication: this is a fix-and-sell or fix-and-re-rate situation, not a long-compounding owner-operator story.

Forensic Red Flags

No SEC exposure (no CIK; regulatory-findings.md returned 0 EDGAR/AAER records). No 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. Item 3 to quote — IQE files UK annual reports, not SEC. Forensic read is therefore from disclosures + structure:

• Going-concern / liquidity risk (the dominant flag, now mitigated): mid-2025 the company put itself up for sale and warned on cash; adjusted net debt rose to £31.5m against a tiny EBITDA base. The £81m May-2026 raise (net £27.9m inflow after debt repayment) mitigated but did not eliminate the structural cash thinness — capex needs will pressure it again.
• Dilution / instrument stack: March-2025 convertible loan notes, then the 2026 stack — 151.5m new shares at 19.8p to MACOM, £15m zero-coupon convertibles (60-month term), warrants over 75.8m shares, plus a £13m placing/retail offer and £23m redemption/reinvestment. Heavy, layered dilution with embedded optionality favouring the strategic investor. Minority shareholders have been repeatedly diluted; the warrant/convertible structure means more potential dilution ahead.
• Adjusted-vs-statutory gap: IQE leans on "adjusted EBITDA" (£3.2m) while the statutory P&L is loss-making (H1 pretax loss £26.5m). Always weigh the statutory loss over the adjusted gloss — the adjustments (restructuring, impairments, site closures) have been recurring, which means they are arguably operating costs, not one-offs.
• Working-capital-flattered cash flow: FY2025 operating cash flow (£8.1m) was boosted by inventory release as the glut unwound — a non-repeatable tailwind. As revenue grows >20% in 2026, working capital will likely consume cash again. Watch for OCF reversing.
• Goodwill/intangibles & impairment cadence: repeated site closures and a dead CMOS++ segment imply impairment risk / a history of write-downs on prior capex — capital that was spent and destroyed.

Regulatory findings (required sub-section):

• SEC (EDGAR LR + AAER): No CIK — no EDGAR search possible. Zero SEC findings by construction.
• Non-SEC web search ("IQE plc" (FTC OR DOJ OR FDA OR settlement OR fine OR penalty) enforcement): no material regulatory enforcement actions surfaced. The only "legal" matter of note was the civil IP dispute with Tower Semiconductor (2022 porous-silicon suit), now amicably settled June 2026 with IQE receiving a royalty-free license — a commercial dispute, not regulatory wrongdoing.
• UK annual-report legal proceedings: not retrieved in full (web-only), but no material litigation flagged in coverage beyond the settled Tower matter.
• Conclusion: No material regulatory or enforcement findings — verified via the empty EDGAR LR/AAER pull and web search as of 2026-06-20. The only litigation (Tower IP) is resolved. The real risk here is financial/dilution, not regulatory/fraud.

---

Phase D — Project & stress-test

Forward Projection

No EPS model is sourceable to research-layer financials (web-only, and IQE is loss-making with a moving share count post-raise), so this is `` with arithmetic, framed as revenue/EBITDA paths, not a false-precision EPS line. IQE's fiscal year = calendar year.

Base inputs:

• FY2025 revenue £97.3m; adj. EBITDA £3.2m (3%).
• FY2026 guide: revenue >+20% YoY; adj. EBITDA high-single to low-double-digit £m.
• ~978m shares o/s pre-warrant/convert dilution.

EPS: n/a — not modellable to a credible figure given (a) loss-making base, (b) shifting share count from warrants/convertibles, (c) no sourced consensus EPS. Stating n/a rather than fabricating an EPS line. The honest forward metric for IQE is EBITDA margin × utilisation × InP-substrate access, gated by capex funding.

Brier forecast: Per --watchlist rules, no our model create in the breadth loop. For the record, the single most scoreable binary: "IQE FY2026 revenue ≥ £117m (i.e. guidance of >+20% met), resolves 2027-03-31" — I'd put p ≈ 0.55. Logged here narratively, not to the tracker.

Bull vs Bear

Bull case. IQE is a scarce, strategically irreplaceable node in the AI-optical supply chain that the market left for dead and is now being re-underwritten by its own customers. The thesis: (1) the open epiwafer market grows ~19% CAGR to ~$2.5bn by 2029 and InP is the bottleneck for 800G/1.6T AI optics — IQE is one of very few merchant players with high-volume InP epitaxy and a non-China footprint; (2) MACOM (11.5% + board + LTSAs) and Tower (multi-year InP minimum-volume deal) have de-risked demand and put real volume commitments behind the InP story; (3) the £81m raise removed the going-concern overhang and gives capacity to invest; (4) defence/mil-sensing is a sticky, funded second leg that re-accelerated in H2 2025; (5) at ~0.9x sales with a fixed cost base, operating leverage is violent on the way up — a return to even 2022-scale revenue at a better mix could swing EBITDA margin to double digits. Contrarian read: the market is still pricing IQE as a failed handset-cycle wafer maker; it is quietly becoming a contracted AI-optical InP supplier with its largest customer underwriting it.

Bear case. Three things could permanently impair the equity (not just the quarter): (1) Chronic sub-scale economics — three regional fabs, a decade of negative ROIC, and capex starvation mean IQE may never earn its cost of capital; "AI optics" could grow the top line while the business still doesn't make money, exactly as 3D-sensing did. (2) Dilution is structural, not episodic — convertibles + warrants + a history of rescue raises mean equity holders get diluted on every cycle; the strategic (MACOM) holds the optionality and the kingmaker board seats, so minorities are price-takers in any eventual sale. (3) InP substrate scarcity cuts both ways — IQE needs to buy InP substrate (AXT/Sumitomo) to sell InP epi; if it can't secure substrate or fund reactors, the AI-optical upside is capped at the source. Pre-mortem (18 months out, thesis broke): FY2026 missed the self-set >20% guide as Wireless kept bleeding and AI-optical ramp slipped on InP-substrate constraints; the working-capital cash tailwind reversed; another raise was needed; MACOM used its board position to take the company private cheap, capping minority upside. Are multiples too high? No — ~0.9x sales is low, appropriately, for the risk; the question isn't the multiple, it's whether any earnings ever materialise. What the market refuses to see (either direction): bulls ignore that "strategically important" and "good equity" are different things (IQE has been the former, never the latter); bears ignore that for the first time in a decade IQE has contracted demand + a recapitalised balance sheet + a focused (shrink-to-InP/defence) strategy.

Devil's Advocate (short-seller)

Dismantling the bull case as a skeptical short:

• What structurally breaks the model: IQE sells a commoditising input (epiwafers) into markets where its customers can in-source or dual-source (Apple did, WIN dominates RF, MACOM could eventually pull epi in-house — note MACOM is now both customer and 11.5% owner, which is a tell that it wants control of the supply, not a healthy arm's-length market). The merchant-epi model has structurally thin margins because the value accrues to the device fab and the substrate maker, not the epi layer in the middle.
• Revenue concentration: Photonics now depends heavily on a narrow set of optical OEMs + the MACOM/Tower relationships. If MACOM's own end-demand disappoints, or it re-sources, a huge chunk of the "growth" story evaporates. The Apple-chain VCSEL business already proved how violently a single end-customer's cycle whips IQE (−66% in 2023).
• Weaker-than-thought moat: epitaxy know-how is real, but WIN Semiconductors, Chinese epi houses (state-backed, Asia = 60% of GaAs), and the substrate makers are all encroaching. The "only full-portfolio tri-regional player" framing is partly a function of IQE being too sub-scale to focus — breadth as a symptom of indiscipline, not a moat.
• Worst capital-allocation history: a decade of dilutive raises and site closures; the 2026 deal structure (zero-coupon converts + warrants to a strategic at 19.8p) is shareholder-unfriendly — it hands cheap optionality and board control to MACOM. Related-party risk: a customer on the board negotiating its own supply price is a structural conflict.
• Assumptions that must hold for today's ~47–57p: (a) FY2026 +20% revenue and margin expansion, (b) InP-substrate access secured, (c) no further dilutive raise, (d) defence funding sustained, (e) MACOM acts as partner not predator. That's five things, several outside IQE's control.
• If growth disappoints 20–30%: at ~0.9x sales the multiple won't cushion much; a revenue miss + a needed raise = the stock round-trips toward the 2025 lows. The asymmetry for a short is the dilution/going-concern recurrence, not the demand story.
• Single scenario that permanently impairs: MACOM takes IQE private at a modest premium using its board seats + shareholding, crystallising the value of the strategic asset for itself and leaving minority holders with a capped, sub-intrinsic exit. Plausibility: moderate-to-high — the company already ran a sale process, and a strategic with 11.5% + two board seats + LTSAs is the natural acquirer.

Management Questions (ordered by information value)

1. InP substrate: Do you have contracted, multi-year InP substrate supply (from AXT/Sumitomo or others) sufficient to meet the MACOM + Tower volume commitments — or is your AI-optical growth gated by an input you don't control?
2. Capex vs capacity: FY2025 capex was only £5.1m. What is the reactor-capacity investment required to hit the bull-case AI-optical volumes, and how is it funded without another dilutive raise?
3. MACOM governance: With MACOM holding ~11.5% and two board seats and being a major customer, how do you ensure InP/GaN supply pricing to MACOM is arm's-length and not value-transfer from minority shareholders?
4. Path to GAAP profit: Adjusted EBITDA is positive but statutory is loss-making with recurring "adjustments." What revenue level and mix takes IQE to sustained statutory operating profit and positive ROIC, and when?
5. Take-private risk: Given you ran a full sale process in 2025, is the board's current intent to operate independently and re-rate, or to deliver a sale? What protects minority holders' upside if MACOM bids?
6. Working capital: FY2025 operating cash flow was flattered by inventory release. As revenue grows >20%, how much cash will working capital consume, and does that re-open a funding gap?
7. Three-fab footprint: Is the tri-regional structure (UK/US/Taiwan) an asset to keep or a cost to rationalise? Post-Taiwan-sale, what is the target manufacturing footprint?
8. Taiwan sale: Status, expected proceeds, and use of funds — and how does losing Taiwan affect Asian customer access and the revenue base?
9. Wireless decline: Is GaAs/GaN Wireless a managed-decline cash cow, a turnaround, or a divestiture candidate? What's the floor on that segment?
10. Defence durability: How much of the Photonics H2-2025 re-acceleration was funded defence programmes vs structural AI-datacentre demand, and how repeatable is the defence piece?
11. Customer concentration: What % of FY2026E Photonics revenue is the top 3 customers (MACOM, Tower, Lumentum), and what is the single-customer concentration risk?
12. Competitive threat: How do you defend InP epi share against Asian state-backed entrants and against customers in-sourcing epitaxy?
13. Dual CEO/CFO role: When does the CEO/CFO role split, and what is the plan to normalise the executive-chair governance structure?
14. GaN power / micro-LED: These got the only growth capex in 2025. What's the realistic revenue timeline and TAM, and are they distractions from the InP main event?
15. 2026 guidance composition: The >20% growth guide — how much is contracted (MACOM/Tower minimum volumes) vs demand-dependent, and what's the downside case if Wireless undershoots again?

---