TSEM Risk Register — Eight Categories

This register codifies the structured-risk view that complements the bear case (narrative form) and the bull case (positive thesis). Each risk is identified by category prefix:

P = Process / operational
C = Customer concentration
K = Capital structure / capex envelope
X = Competitive
G = Geopolitical (Israel-anchored)
R = FPI-regulatory regime
F = Currency / FX
M = Market / macro

Likelihood (L) and Impact (I) are scored 1-3 (1 = low, 2 = medium, 3 = high). The detection trigger field describes the public signal an analyst would observe if the risk were materializing. The mitigation field summarizes management’s commentary or structural defenses where applicable.

The register is calibrated against today’s date (2026-04-29), Tower’s current operating posture (FY 2025 revenue $1.57B ✓, +9% YoY; net income $220M, $1.94 diluted EPS; SiPh revenue $228M = 14.6% of revenue per segment_revenue_mix.md ✓; $920M SiPho/SiGe capex envelope; FY 2028 financial-model target $2.84B / 39-40% GM / 31.7% Op margin per Tower Q4 2025 release ✓), and a spot price of $191.54 / $21.4B market cap (close 2026-04-29).

P — Process / operational risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
P1	PH18 yield-learning curve at next-generation 110 GHz+ tapeouts. Tower’s PH18 is currently a production-grade 1.6T transceiver process at the 200mm node (Tower Q4 2025 release ✓). The Tower-LWLG development agreement (StockTitan ✓) explicitly targets 110 GHz+ for 200G/400G per-lane modulator architectures via “multiple engineering tapeouts during 2026” — the yield-learning at this performance band is a forward observation.	2	3	Joint Tower-LWLG press release of any tapeout failure; Tower Q1-Q4 2026 6-K commentary on PH18-process-yield maturity; LWLG 10-Q + 10-K disclosures (LWLG is US-domestic issuer, richer disclosure cadence).	LWLG has 15+ years of EO-polymer R&D maturity; Tower’s PH18 has multi-customer production-scale-validation. The “multiple tapeouts” cadence creates resilience against any single-tapeout failure event.
P2	Capacity utilization sensitivity — fixed-cost decremental margin. Tower’s gross margin compressed to 20.4% in Q1 2025 (the cyclical floor demonstration per bear case Pillar 1.2) on a 9% sequential revenue compression, implying decremental gross margin of ~25%. The same fixed-cost-leverage works asymmetrically on the upside but exposes severe GM compression in a 2026-2027 specialty-cycle weakness.	2	3	Any future quarter in 2026-2028 where revenue compresses sequentially by >5% AND GM compresses by >150 bps; segment-mix shift toward lower-margin RF Mobile compresses blended GM.	Diversified revenue book (5 named segments: RF Infrastructure 32%, RF Mobile 24%, Power Management 16%, Sensors and Displays 16%, Discretes/Other 12% per segment_revenue_mix.md ✓) reduces concentration impact. Customer-prepayment-funded SiPh capacity (>70% reserved through 2028) provides revenue floor.
P3	Single-fab concentration on Migdal Haemek (geographic-process concentration). Fab 1 (150mm) and Fab 2 (200mm) at Migdal Haemek, Israel (Tower fact sheet ✓) are Tower’s primary Israeli-domestic specialty-process foundries. PH18 silicon-photonics is currently produced at Migdal Haemek; an event-driven disruption (regional conflict, cyber-event, environmental, infrastructure failure) would compress 100% of PH18 revenue to zero immediately.	1	3	Tower 6-K disclosure of operational disruption; Israeli media coverage of regional events affecting Migdal Haemek; satellite imagery of fab activity.	Multi-fab portfolio (Newport Beach US, Agrate Italy, TPSCo Japan) provides partial business-continuity capacity for non-PH18 product lines. PH18 itself is single-source. The diversification mitigates ~60-70% of revenue exposure to a Migdal Haemek event but does not protect PH18 SiPh revenue specifically.
P4	Tower-Intel Fab 11X New Mexico capacity-ramp execution risk. The $300M equipment-investment program at Fab 11X (Albuquerque, NM, acquired from Intel 2024) targets initial wafers H1 2026 productionized H2 2026 ⚠ aggregator-derived. Execution slippage by >12 months pushes out US-domestic 300mm specialty capacity and weakens bull case Pillar 3.	2	2	Q1-Q4 2026 6-K capex disclosures + Fab 11X commissioning announcements; segment-revenue contribution in subsequent 6-Ks.	Intel-built fab infrastructure reduces greenfield-construction risk; Tower has 25+ year operating history with Intel-acquired-fab integrations (Newport Beach Maxim 8” 2016 + Texas Maxim acquisitions). Equipment-vendor (ASML, AMAT, KLA, LRCX, TEL) capacity allocation is the principal critical-path constraint.
P5	Tower-ST Agrate joint venture economics evolution. The Agrate 300mm Lombardy facility shared with STMicroelectronics is a JV where Tower contributes technology + customers and ST contributes facility + capex. JV-restructuring risk (ST renegotiates terms, shifts capacity allocation, alters Tower-share economics) compresses bull case Pillar 2.2.	1	2	ST Microelectronics quarterly earnings commentary on Agrate JV; Tower 6-K disclosures of JV economics; explicit JV-amendment 6-K filings.	Multi-year JV agreement structure with productization milestones (next 65nm BCD productized Dec 23 2024 per Tower release ✓) reduces near-term restructuring probability. ST’s own foundry strategy includes Agrate as a core asset.

C — Customer concentration risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
C1	NTCJ retention is the single largest customer-concentration risk. Nuvoton Technology Corp Japan (NTCJ) is Tower’s largest individually-disclosed customer at 13% of FY 2024 revenue (per FY 2024 20-F acc. 0001178913-25-001537 ✓). NTCJ is a Japanese power-management / image-sensor IDM that operates the TPSCo Japan capacity as a foundry-customer relationship. The TPSCo dependency is structural: TPSCo is Tower’s Japan operations entity and NTCJ is a major TPSCo customer.	2	3	FY 2025 20-F (Apr-May 2026) customer-concentration disclosure shows NTCJ % declines or NTCJ named-customer drops out; NTCJ / Nuvoton Japan earnings calls cite alternate-foundry qualification; Tower 6-K commentary on TPSCo capacity utilization.	Multi-year supply-relationship structure with Japanese-customer-typical multi-year capacity reservations. NTCJ’s product mix (power management, image sensors) is partly captive to TPSCo’s specialized 200mm process. The relationship has been stable for 10+ years post-Tower acquisition of Panasonic-Tower joint operations 2014.
C2	Top 5 customers ~40% of revenue (NTCJ 13% + 4 unnamed at ~27% combined per FY 2024 20-F). The top-4-unnamed-customers concentration (analyst estimate ⚠ inferred at 27%/4 = ~6.5% per customer at the median; could range from one customer at ~12% + others at 4-5%, or evenly distributed at 6-7% each) is structurally significant. Loss of any single top-5 customer would meaningfully compress quarterly revenue.	2	2	FY 2025 20-F customer-concentration footnote shifts; named-customer earnings calls cite migration to GFS, TSMC, SMIC; quarterly revenue prints show step-function declines.	5+ years of customer-relationship continuity at typical Tower customers; sole-sourced design-win structure for many specialty-process products provides switching-cost protection. Aggregator triangulation suggests possible top-5 names: Qorvo, Murata, Broadcom, Semtech, TI, ON Semi, Bosch, STMicro — none of these are individually verified from Tower primary sources; all ⚠ aggregator-only.
C3	Hyperscaler customer concentration in the SiPh segment (post-ramp). Although Tower’s named SiPh customers include Coherent (OFC 2026 demonstration ⚠ specific Tower-PH18 attribution requires verification) and LWLG (development agreement Mar 11 2026 ✓), the revenue concentration is structurally heavy on a small number of hyperscaler-tier or module-OEM-tier customers as the $1B SiPh trajectory materializes. Loss or material reduction of any single Tier-1 photonics customer (Coherent, Marvell, Innolight, Eoptolink) would materially affect the FY 2028 model.	2	3	Q1-Q4 2026 6-Ks + FY 2026 20-F SiPh segment-revenue disclosures show single-customer dependency; named photonics customer (Coherent, Marvell) cites alternate-foundry qualification at GFS Fotonix or Intel SiPh.	Multi-customer roster with active LWLG + Coherent + emerging customers (CPO Foundry Program post-Nov 2025). The customer-prepayment-funded capacity model means customers have committed capital that creates switching-cost protection.
C4	Tier-1 handset envelope-tracker design-win concentration. Per Q4 2025 earnings call slides (Investing.com ✓), Tower disclosed a “previously announced tier-one handset envelope tracker” ramping to 300mm production. Industry triangulation ⚠ inferred suggests this is for Apple iPhone PMIC. Loss of the program (Apple shifts to TSMC / GFS / Samsung Foundry) would compress Power Management +20% YoY growth signal.	1	2	Customer-side commentary (Apple supplier disclosures); Power Management segment growth slowing below 10% YoY in 2026-2027; Tower 6-K commentary on the design-win program.	Apple’s multi-source foundry strategy is structural; sole-source on a single-program-design-win program is rare even at Apple. The design-win is one of multiple Tower power-management programs.
C5	Module-OEM customer (Innolight, Eoptolink, Coherent, Lumentum, Hisense Broadband) consolidation or vertical integration. Module-OEM layer that buys PH18-based components is structurally fragmented; vertical integration (Chinese module OEMs exploring captive SiPh) compresses Tower demand at the module-component level.	2	1	Module-OEM annual reports cite captive SiPh capacity buildouts; module-OEM consolidation events; Tower SiPh-revenue per module-OEM declines.	Tower’s hyperscaler-tier customer relationships (via module integrators) are one architectural layer above module OEMs and structurally less affected.

K — Capital structure / capex envelope risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
K1	$920M SiPho/SiGe capex envelope vs cash balance and free-cash-flow generation. Tower’s $920M commitment (Tower Q4 2025 release ✓) deploys over 2026-2028. FY 2024 YE balance: cash + ST deposits ~$1.22B, total debt ~$0.18B, net cash ~$1.04B ✓. FY 2025 capex was already elevated supporting the early ramp; cumulative 2026-2028 capex including envelope plus base maintenance capex could reach $1.5-2.0B ⚠ inferred. Free-cash-flow generation in 2026-2027 may not fully cover the envelope, requiring (a) cash drawdown, (b) customer-prepayment-funded portion ($920M × 70%+ reserved = $644M+ prepaid), or (c) potential debt issuance.	2	3	Quarterly 6-K cash + ST deposits + total debt disclosures; capex-vs-FCF gap explicitly tracked in 20-F MD&A; explicit debt issuance announcement (8-K-equivalent for FPIs).	Customer prepayments provide structural offset (>70% of envelope reserved with prepayments per Q4 2025 release). Tower’s net cash position provides cushion. The envelope is sized for revenue growth that materially expands FCF generation by 2027-2028.
K2	Debt issuance probability for $920M envelope. If customer-prepayment quantum is below the implied 70% (e.g., 30-50% prepaid), the residual capex requires either cash drawdown or new debt issuance. Tower has historically operated with very low gross debt ($0.18B FY 2024). A $300-500M debt issuance to bridge the capex would meaningfully increase financial leverage.	2	2	Tower 6-K announcements of bond issuance; bank credit facility expansion; 20-F balance sheet showing total debt step-up.	Even at $500M new debt, Tower’s debt/EBITDA would remain reasonable (<2× at projected FY 2027 EBITDA). Investment-grade-equivalent profile likely sustained.
K3	CHIPS Act funding probability + amount. Tower-Intel Fab 11X (Albuquerque, NM) fits CHIPS Act eligibility criteria (US-domestic-fab capacity expansion + specialty-process diversification). Probability ⚠ inferred at 25-40% based on (a) fit, (b) Tower’s documented Fab 11X investment program, (c) post-2026 administration political-economy framing of US-Israeli specialty-foundry partnerships. Probability is upside-skewed: a CHIPS Act award would materially de-risk the Fab 11X capex envelope and improve TSEM equity story. Lack of award is the base case, not a risk per se.	2	2	NIST + Department of Commerce CHIPS Act award announcements; Tower IR press release; 20-F MD&A commentary on CHIPS Act application status.	Tower has not publicly committed to a CHIPS Act application. Even without an award, the Fab 11X program is privately fundable through customer prepayments + cash.
K4	Customer-prepayment quantum disclosure risk. Tower has stated >70% of the $920M envelope is “reserved or in process of being reserved through 2028 with customer prepayments” (Tower Q4 2025 release ✓) but has not separately disclosed the dollar quantum of prepayments. If a future 20-F discloses the prepayment quantum at significantly less than the implied $644M (e.g., $200-300M actual), the capacity-reservation framing weakens and Pillar 3.3 bull case is compromised.	2	2	FY 2025 20-F + FY 2026 20-F balance sheet disclosure of customer prepayments + advance payments line items; Tower IR commentary on the prepayment quantum.	The “reserved or in process of being reserved” language allows for both committed prepayments and contractual capacity commitments. Even at $200-300M actual prepayment, the contractual reservation framing has some validity.

X — Competitive risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
X1	GFS Fotonix at 300mm cost-down threat. GlobalFoundries’ Fotonix is the only 300mm monolithic SiPh process (GFS bull case ✓, GFS SiPh page ✓) operating at ~4× Tower’s revenue scale. The 300mm vs 200mm process advantage compounds: more die per wafer, lower per-die cost, denser monolithic CMOS+photonics integration. As Fotonix capacity scales (AMF acquisition closed Nov 17 2025 ✓ adds 200mm SiPh capacity; Singapore 300mm upgrade trajectory through 2027-2028), the GFS cost-down trajectory threatens Tower’s PH18 ASP power and merchant-SiPh share.	3	3	GFS quarterly SiPh segment revenue prints; OFC 2027 / OFC 2028 customer-roster disclosures; hyperscaler module-OEM commentary citing Fotonix vs PH18 cost economics.	Tower’s PH18 has 5+ years of yield-learning + named-customer roster; the Tower-LWLG EO-polymer integration provides 110 GHz+ bandwidth differentiation that Fotonix does not currently match without comparable polymer integration. This is the load-bearing competitive moat for Tower: process-differentiation at the modulator-bandwidth layer rather than wafer-cost-economics.
X2	Intel SiPh merchant-pivot (post-IDM-2.0). Intel’s IDM-2.0 transformation has been creating partial-merchant-foundry posture. Intel’s SiPh capability includes deeper CMOS+photonics integration than Tower PH18. Risk: Intel converts internal-customer programs (NVIDIA-Mellanox-equivalent, Cisco-internal) plus opens to merchant customers in 2026-2028.	2	2	Intel earnings calls cite SiPh merchant customer wins; module-OEM partner press releases citing Intel SiPh integration; OFC / OCP 2027-2028 customer-roster disclosures.	Intel SiPh has historically struggled with merchant-foundry execution and customer-relationship breadth. Tower’s customer roster represents a structural lock-in advantage. The Tower-Intel Fab 11X relationship (acquired 2024) provides Tower with US-domestic capacity insulation.
X3	TSMC SiPh merchant-launch (post-2027). TSMC has signaled potential merchant-SiPh capacity opening on the 2027-2028 horizon. TSMC’s brand premium with hyperscalers is structural; a TSMC merchant SiPh launch could compress both Tower PH18 and GFS Fotonix share.	2	3	TSMC capex guidance includes SiPh-merchant allocation; OFC / OCP 2027-2028 customer-roster disclosures; hyperscaler module-spec commentary.	TSMC’s existing internal-customer prioritization is a meaningful constraint through 2027. Tower’s 200mm + customer-prepayment-funded capacity model is a different cost-structure layer than TSMC’s leading-edge-comparable economics.
X4	SiCarrier / Chinese specialty-foundry advance under non-US sanctions regime. SiCarrier (China specialty-foundry equipment + process developer per industry reporting; ⚠ aggregator-only) plus advance from SMIC, Hua Hong, Wingtech specialty-foundry programs creates structural Chinese-domestic SiPh capacity that operates outside US export-controlled customer markets but addresses Chinese hyperscaler demand directly.	2	2	SMIC / Hua Hong CY 2026-2028 capex guidance includes SiPh allocation; CIOE / SEMICON China disclosures of Chinese-domestic SiPh customer engagements; Chinese hyperscaler module-OEM commentary.	Tower’s customer book is largely Western (US + EU + Japan) and structurally insulated from Chinese-domestic SiPh competition. The risk is on the Chinese hyperscaler addressable market, which Tower has limited exposure to.
X5	MRVL + Polariton + Celestial AI end-to-end optical stack. Marvell’s $130B-mkt-cap end-to-end positioning (MRVL bull case ✓). Marvell could in principle qualify Polariton’s POH at TSMC SiPh or even at GFS Fotonix instead of Tower PH18 if Tower-Coherent OFC 2026 partnership doesn’t translate to Marvell-Coherent volume reservation on PH18.	2	2	Marvell Q1-Q4 CY 2026 calls; Polariton process-flow disclosures; OFC 2027 disclosures.	Marvell’s existing customer-supplier relationships span multiple foundries. The Tower-Coherent partnership at OFC 2026 + Marvell-Polariton (April 22 2026 close) creates a direct multi-party competitive landscape rather than a clean Tower-loses scenario.

G — Geopolitical (Israel-anchored) risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
G1	Migdal Haemek + Israeli HQ exposure to regional conflict. Tower’s primary fab footprint is Israel: Fab 1 (150mm) + Fab 2 (200mm) at Migdal Haemek (Tower fact sheet ✓). Migdal Haemek is in Northern Israel — within operating range of regional conflict events (Hezbollah / Iran proxy / direct Iranian missile attacks). Tower’s FY 2023 and FY 2024 20-Fs expanded Israel-conflict-risk language post-October 7 2023 Hamas attacks.	2	3	Tower 6-K disclosure of operational disruption (production stoppage, supply-chain interruption, employee call-up to military reserves); Israeli media coverage of regional events affecting Migdal Haemek; satellite imagery of fab activity.	Multi-fab portfolio (Newport Beach US, Agrate Italy, TPSCo Japan) provides 60-70% revenue capacity insulation. PH18 SiPh is single-source at Migdal Haemek and structurally not insulated. Continued Israeli-government Iron Dome / David’s Sling air-defense + diplomatic resolution paths reduce the residual probability.
G2	Israeli sovereign-debt + currency stress under regional conflict. A regional-conflict spike could compress Israeli sovereign credit (NIS depreciation, sovereign-yield expansion) which flows to Tower’s NIS-denominated cost structure (employee compensation, real-estate, local-procurement). Tower reports in USD but pays Israeli operations in NIS.	2	2	Israeli sovereign-bond yields; NIS/USD spot rate; Tower 20-F currency-translation commentary.	Tower has matched-currency hedging program for major NIS-denominated cost categories. NIS-USD volatility historically remains within ±15% bands except in extreme regional-conflict spikes.
G3	Iran-Israel direct military exchange escalation. Direct Iranian missile attacks on Israeli industrial infrastructure (Tower or peer) would trigger insurance + casualty + production-disruption events at meaningful scale. The April 2024 + October 2024 Iranian missile attacks on Israel were militarily-limited but established the precedent.	1	3	Iran-Israel direct missile / drone exchanges; Israeli air-defense engagement events.	Tower has not disclosed direct fab impact from any prior attack event. Israeli industrial-area air-defense coverage is multi-layered.
G4	Regional-conflict-driven employee call-up to military reserves at scale. Israeli reserve-duty call-up affects Tower’s local engineering and operations workforce. October 2023 events triggered material reservist mobilization across Israel; Tower’s employee base includes meaningful reservist count.	2	2	Israeli national mobilization announcements; Tower 6-K commentary on workforce-availability impact.	Tower’s mobilization-resilience structure includes cross-trained operators + automated processes; the operational-disruption coefficient on reserve-duty is small relative to direct fab-disruption events.

R — FPI-regulatory regime risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
R1	HFIAA Section 16 newly applies to FPIs (effective 2026-03-18). The Holding Foreign Companies Accountable Act amendments + SEC FPI rule release (SEC FPI rule release 33-11371 ✓) extend Section 16 short-swing profit recovery + insider-disclosure obligations to FPIs. Tower’s insider class (officers, directors, 10%+ shareholders) is now subject to incremental Form 3/4/5-equivalent disclosure requirements. Implementation friction during 2026 ramp could create unintentional non-compliance events.	2	1	SEC enforcement actions against FPI insiders; Tower 6-K disclosures of FPI compliance ramp; legal-counsel disclosures of compliance program.	Tower has long-standing US securities-law counsel + ADR-equivalent compliance infrastructure. Implementation friction is real but unlikely to produce material enforcement events.
R2	SEC enforcement priorities post-2026-03-18 emphasizing FPI accuracy + customer-concentration disclosure. Recent SEC commentary (post-2026-03-18 rule release) emphasizes FPI customer-concentration disclosure quality + risk-factor specificity. Tower’s customer-concentration disclosure (NTCJ 13% + top-5 ~40%) is already at the disclosure-completeness frontier; future enforcement could push for incremental detail (named top-5 customers, segment-by-customer revenue, etc.).	2	1	SEC comment letters on Tower 20-F; SEC speech / enforcement-priority publications; FPI peer enforcement actions.	Tower’s disclosure completeness is at industry-leading levels; named-customer expansion would be incremental rather than transformative.
R3	CFIUS / national-security review on Tower-related strategic transactions. Tower’s Israeli ownership + US-domestic-fab presence creates structural CFIUS-eligibility for any future strategic transaction. The August 2023 Intel-Tower deal collapse on Chinese antitrust (not CFIUS) doesn’t rule out future CFIUS involvement on a different transaction.	1	2	CFIUS clearance announcements on any future Tower transaction; Tower 6-K disclosures of strategic-transaction discussions.	Tower’s structural alignment with US-allied jurisdictions (Israel + US + Italy + Japan) makes CFIUS-friction unlikely on most transaction types.
R4	Israeli Securities Authority (ISA) regulatory parallel-track risk. Tower is dual-listed on TASE (Tel Aviv Stock Exchange) and subject to ISA regulation in addition to SEC. Israeli regulatory-priority shifts (e.g., Israeli proposed enhanced ESG disclosure, Israeli securities-class-action regime) could create incremental compliance burden.	1	1	ISA regulatory announcements; Tower 6-K commentary on ISA compliance; Tel Aviv-related litigation events.	Multi-jurisdiction compliance is a Tower business-as-usual operation. ISA priorities have historically been more aligned with EU than with SEC, requiring incremental rather than parallel disclosure.

F — Currency / FX risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
F1	NIS/USD exchange rate volatility affecting Israeli operations cost base. Tower reports in USD but pays Israeli operations in NIS. Material NIS appreciation against USD compresses USD-denominated revenue translation; material NIS depreciation reduces USD-denominated cost base (positive). The directional effect depends on whether NIS strengthens or weakens. Historical NIS-USD has ranged from ~3.0 to ~4.0 over 2022-2026 (~33% range).	2	1	Bank of Israel currency-policy announcements; Israeli sovereign-yield spreads; Tower 20-F currency-translation footnote.	Standard FX-hedging program for major cost categories; multi-currency revenue book (USD + EUR + JPY denominated) provides natural hedging.
F2	Israeli central-bank policy 2026 cycle (Bank of Israel rate decisions). Bank of Israel rate decisions affect NIS-USD exchange rate + Israeli sovereign borrowing costs. The 2026 rate-cycle direction is uncertain — Israeli inflation and regional-conflict dynamics drive divergent policy paths.	2	1	Bank of Israel monetary policy committee announcements; Israeli inflation prints; NIS-USD response.	Tower’s currency-hedging is structural; specific Bank of Israel rate decisions affect tactical hedging cost but not strategic positioning.
F3	EUR/USD volatility affecting Tower-ST Agrate JV economics. Tower-ST Agrate operates in EUR-denominated cost base; revenue is partly USD-denominated. EUR-USD movement affects Tower’s share of Agrate JV economics in USD reporting terms.	1	1	ECB monetary policy decisions; EUR-USD spot; Tower 20-F currency footnote.	Standard FX-hedging applies.
F4	JPY/USD volatility affecting TPSCo Japan operations. TPSCo Japan operations are JPY-denominated cost base with multi-currency revenue. JPY weakness compresses USD-translated revenue from Japan operations.	1	1	BOJ monetary policy decisions; JPY-USD spot; Tower 20-F currency footnote + segment-revenue disclosure.	Standard FX-hedging applies. NTCJ relationship (TPSCo’s largest customer) is JPY-denominated, providing matched-currency natural hedge for that revenue.

M — Market / macro risks

ID	Risk	L	I	Detection trigger	Mitigation / management commentary
M1	AI capex digestion 2026-2027. Hyperscaler capex pause / digestion cycle following the 2022-2025 acceleration. Optical interconnect demand softens 6-12 months after compute-capex deceleration. Tower’s PH18 SiPh ramp is exposed; the FY 2028 model assumes sustained AI-capex through 2026-2028.	2	3	Hyperscaler Q4 CY 2025 + Q1-Q4 CY 2026 capex guidance commentary; Big-4 hyperscaler aggregate capex guidance for FY 2026; module-OEM order book guidance.	Multi-customer roster + customer-prepayment-backed reservations through 2028 provide demand floor. The Pillar 1.2 LWLG-PH18 110 GHz+ differentiation creates customer-pull dynamics independent of generic AI-capex digestion.
M2	Mature-node ASP recompression. Chinese mature-node fab buildouts (SMIC, Hua Hong, JCET) add structural overcapacity at exactly the geometries where Tower competes for RF SOI, BCD power, and image-sensor revenue.	3	2	SMIC / Hua Hong CY 2026 capex guidance; merchant-foundry pricing-survey data; Tower segment-revenue prints showing ASP pressure.	Diversified segment mix; specialty-process moats (BCD power, imager pixel, RF SOI) are less commoditized than digital mature-node.
M3	Recession or IT-spend pullback compresses unit-volume demand. Slower hyperscaler + telco infrastructure spend reduces module-OEM order books and PH18 volume conversion. Smartphone unit-volume softness compresses RF Mobile demand.	2	2	Module-OEM order book guidance; carrier capex cuts at telco operators; smartphone-OEM unit-shipment guidance.	AI-driven demand has been more recession-resilient than traditional enterprise IT spend.
M4	Multiple-compression on AI-photonics-narrative re-rating reversal. Tower trades at a meaningful premium to peer specialty foundries (13× EV/Revenue vs VIS 3.2× / DBHiTek 2.0× per comps_valuation.md ✓). If the AI-photonics narrative cools (e.g., FY 2028 model slips), multiple compresses to peer-foundry baseline — implying ~$45/share specialty-foundry-baseline price + AI-photonics premium decay.	2	3	Peer-comp trading multiples; explicit analyst-day re-rating commentary; missed quarterly SiPh trajectory prints.	Premium is justified by the actual SiPh revenue trajectory + customer-prepayment-funded capacity model + Israel-tested resilience track record.
M5	Israeli cost-of-equity premium expansion. Tower’s discount rate embeds an Israel-risk premium that varies with regional-conflict cycle. A spike could compress Tower’s multiple even if operating fundamentals remain intact.	2	2	Israeli sovereign yields; Israeli equity-market risk premium; Tower stock price decoupling from peer-foundry comp set.	Tower has demonstrated resilience through 2023-2024 regional-conflict cycles. The structural-risk-premium is partly priced into the current multiple.

Aggregated risk-scenario sensitivity

Scenario	Risks materializing simultaneously	Implied valuation impact (vs. spot $191.54)
Single-tail bear (most-likely-near-term)	M1 + X1 (AI digestion + GFS Fotonix competitive pressure)	-10 to -15% = ~$163-172
AI-photonics-narrative break combined-stress	M1 + M4 + X1 (AI digestion + multiple compression + GFS Fotonix share)	-25 to -35% = ~$125-145
Customer-loss + cycle-stress	C1 (NTCJ retention) + P2 + M2 (decremental margin + mature-node ASP)	-20 to -30% = ~$135-155
Regional-conflict spike	G1 + G3 + M5 (Migdal Haemek event + cost-of-equity expansion)	-25 to -40% = ~$115-145
Hard-bear combined	M1 + M4 + X1 + C1 + G1 + K2 (multi-pillar bear cascade including debt issuance + regional conflict)	-55 to -75% = ~$50-90 (revisits Intel-deal-anchor $53/sh region)
Single-tail bull (most-likely-near-term)	LWLG-PH18 first tapeout success + Q1 2026 SiPh segment beat	+10 to +20% = ~$210-230
CHIPS Act award + customer-announcement combined	K3 falsified positive + C3 expansion (new SiPh customer named) + multiple-expansion	+25 to +40% = ~$240-270
Multi-bull combined	All three bull pillars compound through 2027-2028 toward FY 2028 model	+35 to +60% = ~$260-310

The hard-bear scenario revisits the Intel-deal-anchor $53/share region (Intel press release ✓) — Tower’s August 2023 floor when the $5.4B all-cash acquisition collapsed on Chinese antitrust. The multi-bull combined scenario reaches the $260-310 range that anchors the bull-case valuation in valuation ranges. The distribution of outcomes is approximately symmetric at current spot — but the bear-tail (regional conflict + AI-narrative break) is mechanically active and should be factored into position sizing.

What this register deliberately does not include

Mubadala-equivalent overhang risks: Tower is fully-distributed-float, has no controlling-shareholder selldown overhang (unlike GFS where Mubadala 77% creates recurring secondary-offering pressure)
CHIPS Act clawback risk (mature US-domestic-foundry concern at GFS): Tower has not received a CHIPS Act award and has no clawback exposure currently. If award materializes, this would become an active risk and would be added at next refresh
DEF 14A proxy / shareholder-vote risks: as an FPI, Tower does not file DEF 14A; AGM mechanics differ; risk is structurally lower-density than US-domestic peers
Cyber-event risk: not separately modeled but implicit in P3 (single-fab concentration); Tower has not disclosed material cyber events. Israeli cyber-defense maturity (national-level) provides structural mitigation