·Dashboard·Research·Work·Archive
← wiki
ticker claude-researchtickerdeep-diveprofilemgmt-ddchecklist updated 2026-04-05

Soitec S.A. (EPA: SOI)

Euronext Paris: SOI | Semiconductors & Semiconductor Equipment | Consolidated April 2026


Identity

What it is: Soitec is the world's dominant supplier of engineered semiconductor substrates -- the specialized wafers at the very bottom of the chip-making stack. Think of it as the foundation of a building: invisible, but everything depends on getting it right.

The core trick: A patented process called Smart Cut works like an atomic scalpel -- implant hydrogen ions into silicon at a precise depth, heat it, and the crystal cleaves along that plane. Bond the ultra-thin crystalline layer onto an insulating substrate, and you get a silicon-on-insulator (SOI) wafer. The result: dramatically better chip performance and energy efficiency.

  • Full name: Soitec S.A.
  • Ticker / Exchange: SOI / Euronext Paris
  • HQ: Bernin, France (near Grenoble)
  • Founded: 1992, spin-off from CEA-Leti
  • IPO: 1999
  • Employees: ~1,991
  • Website: soitec.com
Metric Value (March 2026)
Share price €59.00
Market cap €2.06B
Enterprise value ~€2.19B
52-week range €22.62 -- €59.30
Beta (5Y) 1.44

Segments

Segment What It Does % of FY'25 Revenue
Mobile Communications RF-SOI wafers for smartphone RF filters, FD-SOI for mobile processors. The bread and butter. ~61% (€546M)
Automotive & Industrial FD-SOI and SmartSiC wafers for automotive radar, ADAS, power electronics. ~14% (€129M)
Edge & Cloud AI Photonics-SOI for data center optical transceivers, FD-SOI for edge computing. The star. ~24% (€216M)

Business Model

Pure-play engineered substrate manufacturer. Soitec buys commodity silicon wafers from the big three (Shin-Etsu, SUMCO, GlobalWafers), applies Smart Cut, and sells the resulting engineered substrates to chipmakers at multiples of the raw wafer cost. Revenue is 100% product sales -- no meaningful services or licensing. Revenue is recurring in practice (once a chip is designed on SOI, the foundry needs Soitec's wafers for that product's lifetime) but technically transactional.

Geographic mix: Asia-Pacific accounts for the majority of wafer shipments. Major customers span the US (GlobalFoundries), Europe (STMicro, NXP), South Korea (Samsung), Japan, and China.

Manufacturing Footprint

Facility Location Function Status
Bernin I, II, III France (HQ) Main SOI production (200mm & 300mm); R&D center Operating; world's largest SOI fab
Bernin IV France SmartSiC (silicon carbide) production Inaugurated Oct 2023; ramping slower than planned
Pasir Ris I & II Singapore 300mm SOI production Operating; Singapore II completed 2024, doubling capacity
Hasselt Belgium R&D and small-scale production Operating

Global target: ~4.5M 300mm-equivalent wafers/year by FY'26.

Strategic Partnerships

  • CEA-Leti -- Co-developed Smart Cut. CEA Investissement holds 7.2% equity + board seat. Soitec is essentially the commercial arm of CEA-Leti's substrate innovation.
  • Shin-Etsu Handotai (TYO: 4063) -- World's largest silicon wafer maker; key supplier + board-level partner. SEH holds 0.6% equity.
  • NSIG (National Silicon Industry Group) -- Chinese silicon wafer company holding 5.8%. March 2026: 10-year licensing extension with "no new technology transfer." Politically sensitive given semiconductor export restrictions.
  • Skyworks Solutions -- Multi-year POI supply agreement for Sky5 platform (March 2026).
  • GlobalFoundries -- Key customer, most vocal FD-SOI champion.

Thesis

Soitec is a technology monopoly at a cyclical trough with an AI-driven growth engine the market is not fully pricing. The company controls 70%+ of the global SOI wafer market through 4,300+ patents. No competitor can legally replicate Smart Cut at scale. Meanwhile, Photonics-SOI for AI data centers is accelerating into the strongest secular demand environment in semiconductor history.

Single most important thing that must go right: RF-SOI inventories at customers must normalize within 2-3 quarters, driving Mobile Communications back from ~36% of revenue toward 50%+.

Conviction level: Medium -- exceptional moat and AI optionality, but near-term earnings are genuinely ugly, the stock has already rallied 160% from its low, and it trades above most analyst targets.

Why it matters: As chips get more specialized, the substrate layer -- historically a commodity -- becomes a performance-critical, engineered material. RF filters need RF-SOI for 5G. Optical transceivers need Photonics-SOI to convert electricity to light. EV power electronics need SiC substrates. Soitec sits at this inflection point.

TAM & Market Position

  • SOI wafer market: ~$1.3-1.6B in 2024-25, projected $2.5-3.3B by 2029-33 (7-15% CAGR).
  • Market share: 70%+ of global SOI. No one else has the technology license to produce at scale using Smart Cut. This is one of the strongest positions in all of semiconductors -- comparable to ASML in EUV lithography, though at much smaller scale.
  • Wafer TAM (management frame): ~5M 200mm-eq wafers in 2024 to ~12M by 2030.

Secular Tailwinds

  1. AI infrastructure buildout -- Photonics-SOI demand tracks optical transceiver deployments. Hyperscalers spending $200B+ annually on AI infra. (3-5 year, high magnitude)
  2. 5G proliferation -- More RF filtering content per phone than 4G. POI expanding. (3-5 year, moderate)
  3. Vehicle electrification -- SiC power devices becoming standard in EV powertrains. (5-10 year, depends on SmartSiC execution)
  4. Edge AI / IoT -- FD-SOI's low-power advantage suits always-on edge devices. (3-5 year, moderate)

Business

The Technology -- How Smart Cut Works

The physics: put an insulating layer (SiO2) between active silicon and bulk silicon. You get standard silicon processing compatibility on top with complete electrical isolation below. The challenge was always how to get a perfect crystalline layer bonded to glass.

Smart Cut exploits nuclear physics. Implant hydrogen ions into silicon at a specific depth. The hydrogen accumulates along a plane. Heat it, and the crystal cleaves -- splitting off a perfectly thin crystalline layer. Bond that onto an oxidized handle wafer, and you have an SOI wafer.

[Donor Wafer] → [Ion Implant] → [Bonding] → [Splitting] → [Finishing] → [SOI Wafer]
                  H+ at precise    Face-down     Heat splits    CMP polish,
                  depth (10-100nm) to oxidized    donor along    anneal, QC
                                   handle wafer   implant plane
                                                       ↓
                                                [Donor Recycled up to 10x]

Key engineering challenges: Implant dose uniformity (too much hydrogen = crack; too little = won't split), bond interface quality (any particle = void defect), layer thickness uniformity (±0.5nm across 300mm wafer), surface roughness (<0.1nm RMS).

Key Technical Metrics

Metric State-of-the-Art Why It Matters
SOI layer thickness 5-100nm, ±0.5nm Determines transistor architecture (FD-SOI needs <12nm)
BOX thickness 10-400nm Thicker = better isolation; thinner = better heat dissipation
Defect density <0.05 cm^-2 Defects kill chip yield; lower is better
Wafer diameter 200mm (legacy), 300mm (standard) Larger = more chips per wafer = lower cost
Surface roughness <0.1nm RMS Critical for advanced lithography pattern fidelity
Donor reuse count Up to 10x Direct impact on substrate cost -- massive advantage

Product Deep-Dive

RF-SOI (~61% of FY'25 revenue): Every 5G phone contains RF-SOI in its front-end module for signal filtering. High-resistivity SOI eliminates parasitic coupling at GHz frequencies. ASPs estimated 3-5x commodity silicon wafers. Currently in cyclical correction -- RF-SOI inventories elevated, smartphone growth low single digits.

POI -- Piezoelectric-on-Insulator (within Mobile): Fourth product to reach ~€100M annual revenue. Smart Cut transfers a thin piezoelectric crystal layer onto SOI for 5G BAW filters. 11 customers in volume production, 13+ qualifying. Skyworks Sky5 supply deal validates adoption trajectory.

FD-SOI (across segments): Ultra-thin SOI layer (<12nm) creates a fully depleted channel. Alternative to FinFET for applications where power efficiency trumps raw performance -- automotive radar, IoT, Wi-Fi 7, quantum computing. GlobalFoundries champions 22FDX and 12FDX nodes. Quobly validated 28Si FD-SOI for quantum computing at STMicro's fab.

Photonics-SOI (~24% via Edge & Cloud AI): The star. A 220nm silicon layer on 2μm buried oxide creates an optical waveguide for silicon photonics -- the technology enabling optical interconnects in AI data centers. Every 800G and 1.6T optical transceiver uses Photonics-SOI. Segment grew 27% YoY in Q3 FY'26 while everything else shrank.

SmartSiC (within Auto & Industrial): Smart Cut bonds a thin SiC device layer onto poly-SiC, using 90% less single-crystal SiC than traditional methods (vs. Wolfspeed). Bernin IV facility inaugurated October 2023 but ramp is slower than planned. €41M impairment in H1 FY'26. Diodes Inc. validated SmartSiC (November 2025) but this is the highest-risk product line.

Value Chain Position

[Raw Silicon] → [Silicon Wafers] → [Engineered Substrates] ★ → [Foundries/IDMs] → [End Products]
 (polysilicon)   (Shin-Etsu, SUMCO,   (SOITEC — monopoly)       (GF, STMicro,     (Phones, DCs,
                  GlobalWafers)                                    Samsung)           EVs, IoT)

Soitec takes a $50-100 commodity wafer and transforms it into a $200-500 engineered substrate. Switching costs are extremely high: 12-18 months and millions of dollars to re-qualify at a foundry. Designs specify SOI, locking in the substrate for the product's lifetime.

Competitive Landscape

Company Ticker SOI Relevance Pure-Play?
Soitec EPA: SOI 70%+ share; RF-SOI, FD-SOI, Photonics-SOI, POI, SmartSiC Yes
Shin-Etsu Chemical TYO: 4063 Some SOI wafers at far smaller scale; board relationship No (tiny fraction)
SUMCO TYO: 3436/3436 3436 Minor SOI product line; can't match Smart Cut
GlobalWafers TWSE: 6488 Minimal SOI presence No
Shanghai Simgui Private Growing domestic Chinese alternative; quality gap remains Yes
Wolfspeed NYSE: WOLF SiC substrates; competes with SmartSiC only Yes (SiC)

Competitive Moat

The moat is exceptionally deep, built on four reinforcing layers:

  1. IP / Patents (primary): 4,300+ active patents, 3,000+ covering Smart Cut. No competitor can legally replicate. This is not a "better mousetrap" -- it is closer to a technology monopoly.
  2. Switching costs: 12-18 month re-qualification at foundries. Designs specify SOI, locking in the substrate.
  3. Cost advantage: Smart Cut reuses donor wafers up to 10x. Alternative SOI methods are inferior in quality and cost.
  4. Tacit knowledge: 30+ years of accumulated manufacturing expertise. Cannot be replicated by reading patents.

Moat quality: A+ -- comparable to ASML in EUV lithography.

Porter's Five Forces: New entrants very low (patent wall + capital + qualifications + 30 years of tacit knowledge; Simgui is the only plausible long-term threat). Supplier power moderate (oligopoly, but Shin-Etsu board relationship helps). Buyer power moderate (few alternatives for SOI). Substitutes low-to-moderate (for RF filtering, photonics, and advanced filters, the physics demands an insulating substrate). Rivalry low in SOI (near-monopoly); higher in SiC.

Industry Cycle Position

Where we are: Trough. Revenue has fallen 47% peak-to-trough (FY'23 €1.09B to FY'26E ~€583M). Multiple data points suggest bottoming:

  • Q3 FY'26 revenue €160M was up 18% sequentially (vs. guided mid-to-high single digits -- a meaningful beat)
  • Q4 FY'26 guidance: ~20% sequential growth, implying ~€190M
  • RF-SOI inventory correction appears near its end
  • AI/photonics providing a rising floor

Prior downcycle lasted roughly 2 years. Semiconductor cycles typically run 3-5 years peak to peak.

Emerging Threats

  1. Shanghai Simgui -- Most credible long-term threat. China will close the gap eventually, but Soitec's quality and IP lead should persist for years. The NSIG licensing extension is smart -- controlled technology sharing beats pure competition.
  2. Advanced packaging alternatives -- Some photonics applications could move to heterogeneous integration on standard silicon. 5-10 year threat.
  3. GaN substrates -- For some RF and power applications. Soitec hedging with GaN-on-SOI development (NTU Singapore partnership for 6G, March 2026).
  4. SiC vertical integration -- Wolfspeed, Coherent, Rohm, STMicro integrating SiC substrate production. Threatens SmartSiC specifically.

Management

Executive Team

Name Title Since Background
Laurent Remont CEO Apr 1, 2026 SVP at Infineon (RF & Sensors); 15+ years at STMicro; CTO at Kontron. Engineering from Grenoble INP. This hire looks much better on paper -- someone who actually understands the technology and customer base. Joined March 16, 2026 as advisor for transition.
Pierre Barnabe CEO (departing) Jul 2022 -- Mar 2026 Former EVP at Atos; no semiconductor experience. Outsider hire to replace legendary turnaround CEO Paul Boudre. The executive committee publicly protested his appointment, calling it "incomprehensible." Stock dropped 15% on the news. Departed for "personal reasons" October 2025 (likely partially involuntary -- stock rallied on departure news).
Albin Jacquemont CFO May 2025 30+ years financial leadership: Inetum, Altran/Capgemini, Saur, Darty, Carrefour. Sciences Po + INSEAD. Strong on corporate finance, no semi experience, but acceptable -- you need a CFO who can manage cash in a downturn.
Cyril Menon COO 2006 Soitec lifer. 20 years of institutional knowledge. Ran Bernin HQ 2012-2018. Previously IBM Semiconductor. Harvard Business School. The person who actually keeps the fabs running.
Christophe Maleville CTO 1993 Co-developed Smart Cut applications at CEA-Leti. 30+ papers, ~30 patents. PhD microelectronics, INSEAD MBA. The technology moat personified -- 30+ years of continuous R&D in SOI processing is irreplaceable.
Calvin Chen Acting CSO Dec 2025 Joined 2010; Stanford economics. Leads SOI/FD-SOI development in China.
Steve Babureck CSO Long-tenured President of Soitec USA. Shapes corporate development and IR.
Rene Jonker CPO Recent 20+ years at Philips, NXP, Nexperia, ams-OSRAM.

Board of Directors

Name Role Independent? Background
Frederic Lissalde Chair Yes Former CEO of BorgWarner. Independent with no prior Soitec connections -- a governance improvement over the Meurice era.
Francoise Chombar Director Yes Co-founder/Chair of Melexis (MELE.BR). Genuine semiconductor operating experience.
Samuel Dalens Bpifrance rep No PE/finance. Representing France's sovereign investment arm (11.5%). 12-year tenure -- extremely long, gives Bpifrance deep institutional influence.
Julie Galland CEA rep No Director of Technological Research at CEA. PhD solid-state physics.
Laurence Delpy FSP rep Yes CEO of Kineis. Chairs Compensation and Sustainability.
Satoshi Onishi Director No Former Shin-Etsu Handotai Europe president. 10 years on board.
Christophe Gegout Director Yes Partner at Yotta Capital; former CEA Deputy GM. Chairs Audit.
Shuo Zhang Director Yes CEO Renascia Partners. Deep semiconductor background (Cypress, Altera, Agilent). American. Most globally diversified director.
Maude Portigliatti Director Yes EVP Michelin High-Tech Materials. Materials science PhD.
Delphine Segura Vaylet Director Yes Chairs Compensation.
Victor Barruol Employee Dir N/A In-house lawyer, ethics & compliance.
Didier Landru Employee Dir N/A Senior R&D expert, materials science.

Board composition: 13 members, 64% independent (ex. employee directors), 55% women, 4 nationalities. Met 7 times in FY'25 with 90% attendance. Single-class shares. No poison pill. French double voting rights (2+ year holders) amplify the state block: 45.6M theoretical votes against 35.7M shares.

Governance Dynamics

The most important governance fact: the "Team France" block (Bpifrance 11.5% + CEA 7.2% + FSP 2.5%) collectively controls 6 of 8 seats on the most powerful committees. They share aligned interests in keeping Soitec as a French national champion. When industrial sovereignty and shareholder returns conflict, there is no guarantee the shareholder comes first.

The Meurice era (2019-2024) was governance dysfunction. Former ASML CEO Eric Meurice consolidated power: joined the board in 2018, chaired Nominations, elevated himself to Chairman, pushed through 35 extraordinary resolutions to expand his powers, and engineered the replacement of turnaround CEO Paul Boudre with semiconductor outsider Barnabe. The executive committee publicly protested in an extraordinary letter alleging a "takeover of Soitec by the Chairman." Meurice departed July 2024, replaced by Gegout (interim) then Lissalde.

Governance trajectory is improving. Lissalde is independent with real industry experience. Incoming CEO Remont is a genuine semiconductor veteran. The worst of the board dysfunction appears to be behind the company.

Insider Ownership & Skin in the Game

This is the weakest part of the investment case. Management insider ownership is negligible at 0.02%. No executive has meaningful personal capital at risk.

Name Role Est. Holdings Est. Value How Acquired
Pierre Barnabe CEO (departing) ~1,100 shares ~€80K Open market (230 shares at €94.85 in Sep 2024; additional Nov 2025)
Frederic Lissalde Chair ~1,000 shares ~€73K Open market (1,000 at €91.68, Sep 2024)
Christophe Gegout Audit Chair ~1,250 shares ~€91K Multiple purchases at €83-94 (Sep 2024 -- Nov 2025)
Laurence Delpy Director ~600+ shares ~€44K Multiple at €60-77 (Oct 2024 -- Feb 2025)
Bpifrance Board seat ~4.1M shares ~€300M Strategic investor
CEA Investissement Board seat ~2.6M shares ~€190M Founding investor
NSIG Sunrise Board seat ~2.1M shares ~€153M Strategic investor

Individual director purchases run €20-35K per transaction -- compliance-grade, not conviction-grade. Nobody running this company has meaningful personal capital at risk.

Where directors' money actually lives: Barnabe had an Ipsos board seat (mild attention conflict). Gegout runs Yotta Capital Partners -- an industrial tech investment fund. Zhang sits on three other public company boards and runs/advises three investment funds. Chombar's primary wealth is in Melexis (the semiconductor company she co-founded). For every director, Soitec is a sideshow, not a core position.

NSIG block sale: ~828,000 shares sold May-July 2024 for ~€91M total. NSIG crossed below 10% voting rights. Reads as strategic withdrawal -- not bullish.

Related-Party Transactions

Simgui/NSIG licensing (the big one): Soitec licenses SOI manufacturing technology to Simgui (Shanghai), which produces 200mm SOI wafers in China. Simgui is a subsidiary of NSIG, which holds ~5.8% of Soitec's equity and a board seat. This is a related-party transaction by definition. The March 2026 ten-year extension includes "no new technology transfer" and a "joint commitment to strengthen Soitec's SOI intellectual property rights in China." Terms include expansion of Simgui capacity from 180K to 360K 200mm wafers. Smart move: royalty income from China without deploying capital there. But the EU Commission opened a preliminary inquiry (January 2024) into whether Soitec exchanged commercially sensitive information with its licensees -- this is an active antitrust risk.

Other related-party items: STMicroelectronics MoU (major customer + technology partner; terms not publicly detailed). Former chairman Meurice had a service agreement classified as related-party (terms undisclosed). CEA dual relationship (shareholder + R&D partner) creates subtle conflicts.

Compensation

CEO total comp ~€2M (26% fixed / 74% variable+equity) -- reasonable, arguably low, for a €2.6B market cap semiconductor company. Breakdown: fixed salary €480K, short-term variable €245K (51% of fixed), long-term performance shares ~€1.24M (Onyx 26 plan: 8,637 shares at grant value), pension €14K, benefits €34K.

Performance share plans (Onyx series) use multi-metric hurdles: revenue, EBITDA, EBIT, CSR, relative TSR vs. Euro Stoxx 600 Technology index. Using relative TSR means management cannot benefit from a general market rally without outperforming peers. Including EBIT alongside EBITDA penalizes depreciation-heavy capex cycles. Only 28.3% of Onyx 2024 targets achieved -- these were calibrated for the $2B revenue ambition that was later withdrawn. The hurdles are real, not decorative.

Broad-based Agate 2027 plan covers all ~1,420 employees. This is positive -- equity incentives are distributed broadly, not concentrated exclusively in the C-suite. Three-year vesting schedule with presence conditions.

Capital Allocation Track Record

  • M&A: Disciplined. No major acquisitions. Dolphin Design acquired 2018, divested 2024 in two pieces -- a four-year detour that consumed management attention without adding lasting value. Smart Cut developed in-house.
  • Capex: Mixed. Singapore expansion was sensible (geographic diversification). Bernin IV SmartSiC is the riskier bet -- €41M impairment says it's underperforming. Management responded appropriately by cutting FY'26 capex to €140M from €230M.
  • Buybacks: None despite deeply depressed stock. Defensible (conserve cash in downturn) but no confidence signal.
  • OCEANE convertible bonds: Redeemed October 2025. No convertible overhang. Conversion price was €174 -- most matured without conversion.

Capital allocation grade: C+ -- Adequate but not inspired. Dolphin Design round-trip is a drag. No well-timed buybacks or creative capital deployment.

Credibility & Guidance Tendency

Period Guided Actual Verdict
FY24 Down ~10% €978M, down 10% In line
FY25 Down ~9% (revised) €891M, down 9% In line
Q3 FY26 Mid-to-high single digit sequential % +18% sequential Beat meaningfully
Medium-term $2B revenue, ~40% EBITDA margin Withdrawn May 2025 Overpromised

Straight shooter on near-term guidance (after revisions), overpromiser on medium-term targets. The $2B ambition was cleverly worded -- "ambition" rather than "target" or "commitment" gave maximum flexibility to walk it back. "Reduced visibility" is used repeatedly to justify withdrawing guidance -- defensible but repetitive. "Personal reasons" for the CEO departure is standard French corporate face-saving. None of these rise to deliberate deception, but the pattern shows a management team comfortable with language that preserves optionality.

Overall follow-through rate: ~5/8 trackable commitments (63%). The shift to quarterly-only guidance reduces accountability.

Litigation & Enforcement

  • Former CFO insider trading (2021): Remy Pierre fined €500K by AMF for disclosing inside information (2017-2018). Before current management but reflects imperfect institutional controls.
  • EU antitrust probe (active): European Commission sent formal information requests (January 2024) regarding commercially sensitive info exchanges with licensees (Simgui, Shin-Etsu, GlobalWafers). Preliminary stage but could become material.
  • Patent litigation (resolved): Settled with Silicon Genesis Corp (2017); won against MEMC/SunEdison. Standard IP defense.

Management DD Verdict

Dimension Rating Key Finding
Skin in the Game Red 0.02% insider ownership. Nobody has meaningful personal capital at risk.
Shell / Cross-Holdings Yellow Clean structure. Simgui/NSIG licensing is a legitimate related-party transaction now under EU antitrust scrutiny.
Capital Allocation Yellow SmartSiC capex risk; Dolphin Design detour; no buybacks. Generally sensible.
Compensation Yellow Sound multi-metric structure with real hurdles. But negligible personal ownership limits incentive bite.
Credibility Yellow Near-term reliable. Medium-term $2B target withdrawn. Quarterly-only guidance reduces accountability.
Governance Yellow "Team France" committee dominance. Meurice era was dysfunction. Improving under Lissalde + incoming Remont.
Litigation Yellow Former CFO AMF fine. Active EU antitrust probe. No current executive issues.
Overall Yellow (B-) The moat protects investors despite weak insider alignment. Governance improving.

Green flags: Incoming CEO Remont (genuine semiconductor veteran), CTO Maleville (30 years, 30 patents) and COO Menon (20 years, ex-IBM) provide deep continuity, real performance hurdles (28.3% achieved in downcycle), board refreshment underway.

Red flags: Near-zero executive skin in the game. Board-driven ouster of legendary turnaround CEO Boudre (2022). French state block controls key committees.

Bottom line on management: Not a governance train wreck, but not a company where management alignment inspires confidence. The investment question is primarily about the technology and the cycle. Management is a secondary consideration -- not good enough to be a reason to buy, but not bad enough to be a reason to avoid.


Financials

Fiscal year ends March 31. FY'25 = April 2024 -- March 2025.

Income Statement

Metric FY'23 FY'24 FY'25 FY'26E FY'27E
Revenue €1,089M €978M €891M ~€583M ~€626M
Revenue growth +26% -10% -9% -35% +7%
Gross margin 37.0% 33.9% 32.1% ~25% --
EBITDA €391M €332M €298M ~€154M ~€175M
EBITDA margin 35.9% 33.9% 33.5% ~26% ~28%
EBIT €267M €208M €136M ~-€18M ~€25M
Net income ~€223M €178M €92M ~-€73M ~€29M
EPS (diluted) ~€6.25 €4.88 €2.56 ~-€1.99 ~€0.80

EBITDA margins held up remarkably through the downcycle (33-36%) until FY'26 where volume deleverage, SmartSiC impairments, and mix shift pushed them to the mid-20s. This resilience reflects the monopoly pricing power -- no competitive pressure on price, only volume pressure from end-market weakness.

One-time distortions (H1 FY'26): €41M SmartSiC impairment, €3M Dolphin depreciation, €19M FX loss. Adjusted net income was -€2M vs. reported -€67M.

Cash Flow & Balance Sheet

Metric FY'23 FY'24 FY'25 H1 FY'26 FY'26E
Operating cash flow €262M €165M €202M €26M ~€100M
Capex €186M €177M €172M €56M ~€140M
Free cash flow €77M -€12M €30M -€31M Near zero
Cash position ~€788M €708M €688M €808M* --
Net debt Net cash €142M €34M €94M €145M --
Net debt / EBITDA N/A 0.1x 0.3x 0.5x ~0.9x

*H1 FY'26 cash temporarily elevated pre-OCEANE bond redemption (Oct 1, 2025).

Balance sheet is healthy. €808M cash, €145M net debt. No need to raise equity. Capex sharply reduced (€230M to ~€140M) to preserve FCF through the downturn. Management targeting positive FCF for FY'26.

ROIC

  • FY'23 (peak): ~15-17% -- creating value above ~9-10% WACC.
  • FY'25: ~7-9% -- near WACC.
  • FY'26E (trough): Near zero or negative -- temporarily destroying value.
  • Should recover above WACC once revenue rebounds toward €800M+.

Valuation

Metric Current 5Y Range Peers
EV/EBITDA (TTM) 9.1x 8-25x ASML ~35x, Shin-Etsu ~12x
EV/Revenue (TTM) 2.8x 2-8x ASML ~15x, Shin-Etsu ~4x
P/Book 1.4x 1.4-8x ASML ~20x, Shin-Etsu ~3x
P/E (TTM) 196x 15-50x (normalized) N/M at trough

Trading at the low end of its own historical range on every metric. At 1.4x book, you are paying barely above liquidation value for a technology monopoly. The headline P/E of 196x is meaningless -- earnings are at cyclical trough. EV/EBITDA of 9x on trough EBITDA is the useful gauge.

What the market is pricing: Revenue stabilizing at €600-650M (40% below peak), EBITDA margins recovering to ~28-30%, modest growth. This feels conservative for a monopoly supplier with strong secular tailwinds.

Scenario Analysis (FY'28E)

Scenario Revenue EBITDA Margin EBITDA EV/EBITDA Per Share
Bear €650M 28% €182M 10x €48
Base €850M 32% €272M 12x €89
Bull €1,050M 35% €368M 14x €142

Simple DCF sanity check: Revenue recovers to €850M by FY'28, grows 5% after; 33% EBITDA margin; 15% capex/revenue; 9.5% WACC; 3% terminal growth. Fair value: ~€80-90 per share.

Incremental Margin Analysis

Sequential revenue ramp (€92M → €139M → €160M → ~€190M) is encouraging. Fixed cost base is already in place. On the way back up, Soitec should convert incremental revenue at 40%+ EBITDA margins given Singapore expansion is complete and capex is declining.

Ownership

Holder Type %
Bpifrance Sovereign/Strategic 11.5%
Baillie Gifford Institutional (Active) 11.3%
BlackRock Institutional 8.9%
CEA Investissement Strategic (founding) 7.2%
NSIG Sunrise Strategic/Financial 5.8%
GIC Pte Ltd Sovereign (Singapore) 5.0%
Point72 (Steve Cohen) Hedge Fund 5.0%
FSP French institutional 2.5%
Employees Insider 1.4%
Management Insider 0.02%

Baillie Gifford at 11.3% is a strong signal -- they are among the world's most respected long-term growth investors. Their presence suggests a compounding story, not a trade. Point72 at 5.0% is a more catalyst-oriented bet, likely on the cyclical recovery. Total institutional: ~48%.

Share buybacks: Count declined 2.6% YoY -- retiring shares, not diluting.

Analyst Sentiment (March 2026)

Metric Value
Consensus Accumulate (18 analysts)
Avg price target €35.78
Range €22 -- €55
Current price €59.00 (69% above avg target)

Major disconnect. Either analysts are lagging (targets set at lower prices, not updated for AI/photonics momentum) or the market is overshooting (rally from €22 to €59 ahead of fundamentals). Probably both.


Catalysts & Risks

Near-Term Catalysts (0-12 months)

  1. CEO Remont takes the chair (April 1, 2026) -- semiconductor veteran; market should welcome.
  2. FY'26 full-year results (May 2026) -- if FCF is positive as guided and FY'27 guidance shows recovery, re-rating potential.
  3. RF-SOI inventory normalization -- when customer inventories clear, orders should snap back.
  4. Additional POI supply agreements -- Skyworks deal is the template.

Medium-Term Catalysts (1-3 years)

  1. Photonics-SOI volume inflection -- 1.6T transceivers and co-packaged optics drive step-function demand.
  2. FD-SOI foundry expansion -- if Samsung or TSMC add FD-SOI nodes (beyond GlobalFoundries), addressable market expands significantly.
  3. SmartSiC qualification wins -- 2-3 major foundry qualifications de-risk the SiC bet.
  4. GaN-on-SOI for 6G -- NTU Singapore partnership could open a new substrate market.

Technology Roadmap

  • Next-gen SOI: thinner layers (<10nm), lower defect density (<0.05 cm^-2)
  • Advanced POI for next-gen 5G/6G acoustic filters
  • SmartSiC cost reduction via improved donor SiC reuse
  • 28Si FD-SOI for quantum computing (Quobly validation at STMicro)
  • 18 new patent filings in 2024-25; 4,300+ active patents

Risk Matrix

Risk Likelihood Impact Mitigant
RF-SOI correction extends into FY'28 Medium High Q3 FY'26 sequential improvement suggests bottoming. But if inventories take 2-3 more quarters, stock could re-test €35-40.
SmartSiC write-down -- qualifications stall Medium-High Medium Even full write-off doesn't threaten balance sheet (net debt 0.5x EBITDA). Core SOI business unaffected. €41M already impaired.
Chinese competition -- Simgui achieves quality parity Low (near-term), Medium (5-10yr) High 4,300+ patent wall, quality gap, 12-18 month qualification barrier. NSIG licensing for controlled technology sharing.
Customer concentration Medium Medium Inherent to semi supply chain. Mitigated by monopoly -- customers need Soitec more than vice versa.
EU antitrust probe escalates Low-Medium Medium At preliminary inquiry stage. Could be material if Commission finds anticompetitive coordination in licensing arrangements.
Photonics-SOI demand plateaus Low High Alternative SiPho integration methods are 5-10 years out. Near-term demand trajectory is strong.

Bear Case

If the cycle doesn't recover and SmartSiC is written off, the stock trades back to €35 (1.0x book value). Maximum drawdown from current price: ~40%.

Thesis Invalidation Triggers

  • FY'27 revenue below €550M (no recovery from trough)
  • Major foundry de-qualifies SOI substrates for alternative technology
  • Shanghai Simgui achieves competitive quality and wins domestic foundry qualifications at scale
  • Loss of a major customer

Decision Log

2026-04-07 — Asymmetrical Bets thesis (Substack, Feb 25 post)

External validation of the SiPho thesis from Asymmetrical Bets after 36 hours at Nvidia GTC studying networking & CPO:

"Soitec at €54 / $SLOIF / $SOI.PA is the clearest photonics multi-bagger we see right now. Now my third largest asymmetrical bet, behind AAOI and KRKNF. Steel conviction. Soitec owns 95%+ market share in Photonics-SOI wafer, the substrate underneath every silicon photonics chip. Every transceiver from AAOI and COHR. Every CPO switch from Nvidia, Broadcom, Marvell. All fabricated on Soitec substrates. Their only licensed competitor, GlobalWafers, is exiting. There's an incredible moat here. No one else is qualified at all the foundries and it would take competitors years to even try and insert at TSM, TSEM, GFS as they ramp for SiPho. Why it's asymmetrical beyond the market growth: co-packaged optics requires 4x the SiPho wafer content vs pluggables."

Datapoint reinforces the entry-on-pullback thesis. Doesn't change the WATCH stance — €59 is still over-extended — but the moat and CPO content multiplier are exactly the reasons to be patient and scale in on a pullback toward €48-52 rather than capitulate on the watch.

Current Recommendation: WATCH -- Scale-In on Pullback

(As of March 14, 2026 at €59.00)

Soitec is a genuinely exceptional business. The moat is A+ quality, the cyclical setup is attractive, and the AI tailwind is real. Recent catalysts validate the story -- Skyworks POI supply deal (March 5), NSIG licensing extension (March 13), incoming CEO Remont (semiconductor veteran). But the stock has rallied 160% from its 52-week low and 43% in the last week alone. It trades above all analyst targets and is deeply overbought on every timeframe.

Do not chase this move at €59. The quality of the business is not in question -- the question is purely about entry price and timing.

Entry Strategy

Conviction: Medium | Target position: 2-3% of portfolio | Scale in over 3-6 months

Tranche Price Size Rationale
1 €48-52 1.0% Pullback to 50-day MA support
2 €42-45 0.75% Retest of breakout level
3 €55-58 0.50% Confirmed consolidation (if tranche 1 not triggered)

Exit criteria:

  • Sell 50% at €80, remainder at €100+ (base case implies €89 fair value by FY'28)
  • Exit fully if FY'27 revenue below €550M
  • Time stop: reassess if no revenue inflection by Q2 FY'28

Technical Context (March 14, 2026)

  • Stock has broken out of year-long downtrend, forming massive rounding bottom from €22.62 low
  • RSI likely 85-90 -- deeply overbought
  • 50-day MA ~€32; 200-day MA ~€37 -- golden cross forming
  • Volume spike (1.18M shares Mar 12) confirms institutional participation
  • Technical verdict: Do NOT buy now. Wait for pullback.

Support levels: S1 €48-50 (gap-up zone), S2 €42 (volume support / 50-day convergence), S3 €35 (200-day MA / deep pullback) Resistance: R1 €59-60 (52-week high zone), R2 €70, R3 €80 (base case fair value)

Key Dates

  • April 1, 2026 -- CEO Remont officially takes the chair
  • May 27, 2026 -- FY'26 full-year results (revenue will look ugly, but trajectory and FY'27 guidance matter more)
  • Q1 FY'27 -- First quarter under new CEO; watch for strategic direction signals

Behavioral Traps Audit

  • FOMO risk: Stock rallied 160%. But still 70% below all-time high of ~€200. Set limit orders and wait.
  • Confirmation bias: Technology moat story is compelling -- easy to rationalize buying without accounting for near-term earnings reality. Must weight the bear case.
  • Narrative seduction: "AI data center + technology monopoly" is extremely attractive. Verify Photonics-SOI revenue is actually material enough to offset Mobile and Auto declines. Current data: Edge & Cloud AI was €216M in FY'25 -- meaningful but not enough to carry the whole company alone.

Proceed with caution -- size smaller and scale in rather than going all-in.


Recent Developments (as of March 2026)

Q3 FY'26 revenue (Feb 3, 2026): €160M, up 18% sequentially -- meaningfully beat guided mid-to-high single-digit growth. Mobile €90M, Edge & Cloud AI €54M (up 27% YoY at constant FX), Auto & Industrial €16M. 9M FY'26: €390M, down 26% YoY. Q4'26 guided ~20% sequential growth.

Key takeaways from recent quarters:

  1. AI is the bright spot. Photonics-SOI is the only growing segment. Edge & Cloud AI grew 27% YoY in Q3 at constant FX. Management described momentum as "very favorable."
  2. Mobile is bottoming. RF-SOI inventory correction appears near its end. Q3 mobile revenue sequentially improved from Q2 levels.
  3. Automotive is ugly. Down 61% YoY for 9 months. Global automotive chip correction is deep, SmartSiC ramp is delayed.
  4. SmartSiC impairment. €41M in H1 -- management acknowledged SiC market ramp is "longer than initially anticipated." Customer qualification cycles are slow.
  5. Guidance withdrawn to quarterly-only. Unusual for Soitec. Reflects genuine uncertainty.

March 2026 catalysts:

  • Skyworks multi-year POI supply deal (Mar 5) -- validates POI adoption trajectory
  • NSIG 10-year licensing extension (Mar 13) -- "no new technology transfer"; subject to NSIG shareholder approval
  • NTU Singapore GaN-on-SOI partnership -- targeting 6G applications
  • CEO successor Laurent Remont announced; joins as special advisor Mar 16, effective Apr 1

Next earnings: FY'26 full-year results expected May 27, 2026. Revenue will look ugly, but the trajectory and FY'27 guidance matter more.


Sources

Topics