Soitec
B2B sale of engineered wafers (RF-SOI, FD-SOI, Photonics-SOI, Imager-SOI, POI, SmartSiC) to foundries and IDMs, plus Smart Cut technology licensing royalties; capex-heavy fabs in France, Singapore and Belgium; revenue largely USD-denominated with EUR cost base
Earnings, margins, COGS & capex
FY26 (ended March 2026) was the trough of a violent down-cycle: revenue fell 34% to EUR 592M as the RF-SOI smartphone inventory correction (Mobile Communications -41% cc to EUR 309M) and automotive weakness (-44% cc to EUR 69M) overwhelmed 8% cc growth in Edge & Cloud AI (EUR 214M, +19% ex-Imager-SOI). Gross margin halved to 16.3% on low fab loading and the company swung to a EUR 220M net loss (heavy depreciation on top of the operating loss; EBITDA stayed positive at EUR 151M / 25.4%). Cash discipline was the bright spot: capex cut 41% to EUR 135M, FCF back to +EUR 63M, net debt just 0.4x EBITDA. Photonics-SOI crossed $100M revenue earlier than planned and Q4'26 (EUR 202M, +25% QoQ cc) beat the ~+20% sequential guidance, marking a probable cyclical bottom.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~84¢ is cost of goods and ~16¢ operating expense, and the remainder is an operating loss.
Revenue trend
Margins
collapsed from 32.1% FY25 (EUR 286M) on under-utilization
down from 33.5% FY25 (EUR 298M) but resilient vs revenue drop
swung from +EUR 119M (~13%) FY25; D&A-heavy model punishes low volumes
vs EUR 92M profit FY25
positive inflection vs -EUR 23M FY25 on capex cuts and working-capital discipline
COGS structure
COGS dominated by raw polished silicon wafers (Shin-Etsu, SUMCO), fab depreciation, energy and cleanroom labor across Bernin (France), Singapore and Hasselt (Belgium) sites; fixed-cost-heavy structure means utilization swings drive violent gross-margin torque in both directions (32.1% -> 16.3% in one year on the way down).
Capex
EUR 135M FY26 (-41% YoY) after a heavy multi-year 300mm capacity build; guided ~EUR 100M FY27. Prior-cycle expansion left substantial installed capacity, so the photonics/FD-SOI ramp can be served with low incremental capex -- positive for FCF if volumes return.
Latest earnings
Beat the ~+20% cc sequential Q4 guidance and the EUR 584M FY26 revenue consensus; FCF of EUR 63M crushed the ~EUR 6M consensus -- shares surged ~14% to their highest since August 2023 on the print
Q1 FY27 revenue expected up ~15% YoY at constant currency and scope; FY27 capex ~EUR 100M; management flags continued profitability pressure from low fab utilization, FX headwinds and reduced government funding
- FY26 revenue
- EUR 592M (-34% reported)
- FY26 EBITDA
- EUR 151M (25.4% margin)
- FY26 net result
- -EUR 220M
- FY26 FCF
- +EUR 63M (vs -EUR 23M FY25)
- Net debt / EBITDA
- 0.4x (net debt EUR 56M, cash EUR 562M)
- Photonics-SOI revenue
- >$100M in FY26, earlier than initially anticipated
Growth drivers
- Photonics-SOI for AI datacenter optical interconnects — >$100M revenue in FY26, earlier than initially anticipated, penetrating high-speed pluggable transceivers with growing volumes shipped for co-packaged optics (CPO) industry qualifications (first explicit CPO confirmation from management, per Bernstein); Morgan Stanley back-calculates ~$130M and 60-70% FY26 growth
- Optical interconnect market projected from ~$14B to ~$73B by 2030, with silicon-photonics penetration of AI optical units rising from ~38% to ~84% (Bank of America optical-interconnects primer, March 2026, which also puts Soitec at >95% share of photonics-grade SOI)
- FD-SOI wafers for edge-AI, IoT and automotive compute (STMicroelectronics, GlobalFoundries platforms)
- Cyclical RF-SOI recovery as smartphone-chain inventories normalize (client wafer inventories ~2M units at March 2026, down from ~2.5M in June 2025 but flat since December 2025)
- POI (piezoelectric-on-insulator) for RF filters and SmartSiC for EV power electronics as longer-dated options
Bull & bear
A depressed, FCF-positive near-monopoly at the cycle trough with a free AI option that is already monetizing: Photonics-SOI passed $100M earlier than planned, CPO qualification volumes are shipping before the ramp, and RF-SOI inventories are past peak -- revenue can rebuild toward EUR 900M+ with 30%+ EBITDA margins on capacity that is already built and largely depreciated.
- Photonics-SOI >$100M in FY26 (Morgan Stanley back-calculates ~$130M, +60-70%) while the optical interconnect TAM compounds from ~$14B to ~$73B by 2030 with silicon photonics taking share -- Soitec is the substrate gatekeeper (>95% photonics-grade SOI share per BofA) for that transition
- Operating leverage in reverse: the same fixed-cost fabs that produced a 16.3% gross margin at EUR 592M revenue produced ~32% at EUR 891M (FY25) and record profitability at the EUR 1,089M FY23 peak -- a volume recovery mechanically restores profitability with only ~EUR 100M/yr capex
- Cycle bottom is dated and visible: Q4'26 +25% QoQ cc beat the ~+20% guide, Q1'27 guided +15% YoY cc, client RF-SOI inventories down from ~2.5M to ~2.0M wafers
- Balance sheet de-risks the wait: EUR 562M cash, 0.4x net debt/EBITDA, FCF positive even in the loss year
- Moat is structural: Smart Cut IP plus 20+ year foundry qualifications mean competitors compete for the minority of the market Soitec does not already hold; strategic EU backing limits hostile outcomes
- Fresh CEO (Laurent Remont, ex-Infineon RF-and-sensors chief, from 2026-04-01) with a mandate to pivot the portfolio toward AI/photonics gives a re-rating catalyst on a credible strategy update
A structurally challenged, customer-concentrated wafer supplier whose core RF-SOI franchise may never recover to peak, wearing an AI-photonics multiple: at EUR 3.77B market cap the stock trades near ~6.5x trough sales and ~25x trough EBITDA while losing EUR 220M net -- priced for a CPO ramp that is still in qualification, against rising substrate competition and Chinese localization.
- The core is shrinking, not pausing: Mobile Communications fell 41% cc and RF-SOI client inventories have been stuck at ~2M wafers since December 2025 -- flat, not falling -- while RF content shifts and bulk-CMOS alternatives cap the recovery
- FY26 P&L damage was severe (gross margin 16.3%, EUR 131M operating loss, EUR 220M net loss) and management itself guides continued profitability pressure from low utilization, FX and reduced government funding
- Photonics is real but small: >$100M is roughly 15% of revenue; even strong growth cannot offset the ~EUR 300M of revenue lost since FY25 for several years -- Jefferies models only EUR 630M FY27 revenue and sees no photonics acceleration before FY30
- Competition is arriving exactly where the growth is: GlobalWafers (CHIPS-backed US SOI capacity), Shin-Etsu/SUMCO scale advantages, and NSIG/Simgui domestic Chinese capability threaten both share and pricing
- CPO timing risk: qualification volumes are not production ramps; Jefferies explicitly sees limited CPO adoption in scale-up before ~FY30, and the stock's 52-week ~9x range (EUR 22.62-200.50) shows how brutally the narrative reprices -- the downgrade alone took shares down ~13%
- Governance/execution overhang: CEO transition mid-crisis, guidance credibility damaged by the FY25-26 warnings cycle
What it is worth
EV/Sales and EV/EBITDA on trough figures, cross-checked against a sum-of-the-parts (legacy SOI franchise at wafer-peer multiples + Photonics-SOI at AI-optics growth multiples); market cap EUR 3.77B (~$4.2-4.4B), EV ~EUR 3.83B (net debt EUR 56M)
Mobile RF-SOI recovery stalls (inventories stuck at ~2M wafers), CPO ramp slips toward FY30 as Jefferies expects, GlobalWafers/China capacity pressures price -> revenue stuck near EUR 600-650M with continued net losses; derating toward wafer-peer multiples implies the EUR 55-85 zone (Jefferies PT EUR 85, street low EUR 55)
Gradual FY27-28 recovery (Q1'27 +15% cc guide extends; sell-side FY27 estimates cluster around EUR 630-700M, FY28 consensus ~EUR 825M), gross margin rebuilds toward mid-20s%, photonics grows double digits -> stock range-bound around EUR 90-140 (analyst consensus target ~EUR 137) pending CPO proof points
Photonics-SOI scales multi-fold as CPO ramps, mobile recovers toward prior run-rates, EBITDA margin back above 30% -> revenue back toward the ~EUR 1B FY23-peak zone, re-rating toward the recent EUR 200 high
At ~6.5x FY26 EV/Sales and ~25x FY26 EV/EBITDA the stock is priced well above silicon-wafer peers -- Jefferies puts it at ~34x FY27 EV/EBITDA, an ~85% premium to the ~18.3x peer average, and cut to underperform (PT EUR 85). The premium is the AI-photonics/CPO option plus cyclical-recovery torque. Multiples on trough earnings overstate expensiveness: on a recovered ~EUR 900M revenue / ~30% EBITDA scenario, EV/EBITDA compresses to ~14x. The 52-week EUR 22.62-200.50 range shows the market has priced both the bear (wafer-commodity derating) and the bull (photonics monopoly) within one year. Not financial advice.
SWOT
Strengths
- Near-monopoly in engineered SOI substrates (~70%+ of the SOI wafer market as widely cited; >95% of photonics-grade SOI per Bank of America, March 2026), anchored by the patented Smart Cut layer-transfer process and decades of process know-how
- Photonics-SOI is the substrate of record for silicon photonics — direct leverage to AI optical interconnect and CPO buildout, already >$100M revenue
- Robust balance sheet through the trough — EUR 562M cash, net debt 0.4x EBITDA, FCF positive at the bottom of the cycle
- Deep qualification lock-in with foundries/IDMs (GlobalFoundries, STMicroelectronics, Tower Semiconductor, UMC, Samsung) -- substrates are spec'd into multi-year platforms
- EU strategic-asset status with French state-linked anchor shareholders (Bpifrance, CEA) and France 2030 / EU Chips Act support
Weaknesses
- Extreme end-market concentration — Mobile Communications (RF-SOI) was still ~52% of FY26 revenue and fell 41% cc -- a one-cycle, few-customers exposure
- Fixed-cost-heavy fab model — gross margin halved to 16.3% and the company swung to a EUR 220M net loss on volume decline
- FY26 loss and low utilization to persist near term — management itself flags FX and shrinking government-funding tailwinds
- Leadership churn — CEO Pierre Barnabe announced his resignation 2025-10-01; successor Laurent Remont (ex-Infineon SVP for RF and sensors, ex-STMicroelectronics) only took over 2026-04-01 -- strategy under a new chief unproven at Soitec
- Share-price hyper-volatility (EUR 22.62-200.50 in 52 weeks) reflects a narrative-driven register, complicating capital raising and M&A currency
Opportunities
- CPO ramp at hyperscale AI clusters — qualification volumes already shipping; each CPO-generation switch/accelerator embeds far more photonics content per system
- Silicon-photonics penetration of AI optics from ~38% to ~84% by 2030 (BofA primer) expands the served market multiples beyond today's transceiver niche
- RF-SOI cyclical snap-back — inventories down from ~2.5M (June 2025) to ~2M wafers; any smartphone-content recovery restores high-margin volume on already-depreciated capacity
- FD-SOI adoption in edge-AI/automotive (22FDX/18FDX-class platforms) and POI in next-gen RF filters
- Smart Cut licensing/royalty expansion into new material systems (SiC, piezoelectrics, compound semis)
Threats
- Substrate substitution — bulk-CMOS RF integration, silicon-nitride and InP photonics platforms, or glass/organic interposers could bypass SOI in parts of the roadmap
- Competitor capacity — GlobalWafers adding SOI capacity (incl. CHIPS-Act-backed US expansion at its Missouri MEMC site), Shin-Etsu and SUMCO able to push SOI lines; China's NSIG/Simgui building domestic capability beyond its Soitec contract
- China localization — mainland self-supply of RF-SOI substrates would erode the largest single demand pool (analysis context only)
- Customer concentration risk — a GlobalFoundries or STMicroelectronics platform shift or insourcing decision moves the P&L materially
- Valuation/narrative risk — stock already re-rated hard on AI photonics -- Jefferies cut to underperform (PT EUR 85) on overvaluation, seeing no Photonics-SOI acceleration before FY30 on limited CPO adoption in scale-up, and shares fell ~13%; any CPO ramp delay hits a rich multiple
Moats, dependencies & bottlenecks
Moats
Smart Cut patented layer-transfer process + accumulated process know-how high on know-how even as early patents age; extended into POI/SmartSiC The core reason most SOI wafers are Soitec's; licensing (historically to Shin-Etsu, Simgui) is controlled by Soitec
high per platform generation Substrates are qualified into foundry platforms (GF RF-SOI/FD-SOI, ST FD-SOI, photonics PDKs) for multi-year lifecycles; requalification is costly and risky for customers
Pre-built capacity gives cost advantage in an upturn but is the source of the utilization-driven losses in the downturn
contested by SiN/InP platforms and potential new SOI entrants Substrate of record for transceiver and CPO qualifications; >95% share of photonics-grade SOI per BofA (March 2026)
Dependencies
GF is the flagship RF-SOI customer; client inventories (~2M wafers, 8-inch equivalent) and smartphone RF content decisions drive the largest revenue block
GlobalFoundries, Tower Semiconductor) Platform-level design wins concentrate demand in a handful of foundry decisions
Base-wafer input from two Japanese suppliers who are also potential SOI competitors
Simgui produces 200mm wafers contractually sold through Soitec; also the seed of a future domestic Chinese competitor (context, not a call)
Management explicitly flags reduced government funding as an FY27 profitability headwind
USD revenue vs EUR cost base; FX drove the gap between -34% reported and -30% cc FY26 revenue and is flagged again for FY27
Advantages
- Dominant share of a niche it created — engineered SOI substrates -- with the only proven high-volume layer-transfer process (Smart Cut)
- Direct, already-monetizing exposure to AI optical interconnect via Photonics-SOI (>$100M, earlier than planned) without competing against its own customers
- Trough-tested financial model: positive EBITDA (25.4%) and FCF (EUR 63M) in a -34% revenue year
- Capacity already built: recovery volumes carry low incremental capex (~EUR 100M/yr guided)
- European strategic-asset positioning: policy support and protection from hostile acquisition
Weaknesses
- Single-cycle concentration: smartphone RF-SOI still >50% of revenue
- Utilization-sensitive cost structure produces outsized losses in downturns (EUR 220M FY26 net loss)
- Narrow direct-customer base (a handful of foundries/IDMs)
- New CEO with no track record at Soitec (industry background at Infineon/STMicroelectronics); guidance credibility still rebuilding after the warnings cycle
- Equity extremely volatile (EUR 22.62-200.50 over 52 weeks), reflecting a narrative-driven shareholder register
Bottlenecks
- Fab utilization — fixed-cost absorption is the single biggest P&L lever; gross margin cannot normalize until volumes recover
- CPO industry ramp timing — Soitec ships qualification volumes but revenue scale awaits hyperscaler/switch-vendor production schedules it does not control
- RF-SOI inventory digestion stalled at ~2M wafers since December 2025 -- the recovery clock has not restarted in mobile
- Customer platform decisions (few foundries) gate all substrate demand; Soitec has no direct route to end markets
Top signals & trends
Top signals
AI photonics revenue is real and pre-CPO-ramp; Morgan Stanley back-calculates ~$130M and 60-70% growth
Positions Soitec in the next optics architecture before the ramp
FCF of EUR 63M vs ~EUR 6M consensus was the key beat; cycle trough likely passed
First YoY-growth quarter guide after the slide from the FY23 peak (EUR 1,089M) to EUR 592M
Mobile destocking stalled; core recovery not yet underway
P&L repair lags the revenue turn
Sell-side pushback on the AI re-rating
Reset opportunity but execution unproven at Soitec; watch first strategy update; his board seat goes to the July 2026 AGM
No balance-sheet distress; company can wait for the cycle
Trends
Market projected ~$14B -> ~$73B by 2030 (BofA optical-interconnects primer, March 2026); silicon photonics on SOI is the dominant platform
Directly expands Photonics-SOI substrate demand
The FY25-26 bust; recovery timing uncertain with inventories stalled at ~2M wafers
EU support helps Soitec; CHIPS-backed GlobalWafers US SOI capacity and Chinese self-supply threaten share
Automotive & Industrial fell 44% cc in FY26; SmartSiC monetization pushed out
FD-SOI demand driver cited in the FY27 outlook
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Raw polished silicon base wafers
Raw polished silicon base wafers
Contract 200mm wafer production sold through Soitec
Wafer-bonding equipment central to layer transfer
Flagship RF-SOI and FD-SOI foundry customer; epicenter of the inventory correction
FD-SOI IDM partner and customer
RF-SOI and silicon-photonics foundry customer
Silicon-photonics/CPO platform builder using SOI substrates
RF-SOI foundry customer
FD-SOI platform licensee/customer
RF front-end module maker downstream of RF-SOI wafers
World's largest silicon wafer maker; historical Smart Cut licensee with its own SOI capability
No.2 silicon wafer maker; SOI product line, scale cost advantage in base wafers
Expanding SOI capacity including CHIPS-Act-supported US sites (Sherman, Texas 300mm; MEMC St. Peters, Missouri incl. SOI) -- the most direct new capacity threat
Korean wafer maker with SOI ambitions and captive Korean customer base
China's domestic SOI champion; currently sells 200mm output through Soitec by contract but is the vehicle for Chinese SOI localization
Specialty SOI/sensor wafers at 200mm
Not wafer competitors but alternative photonics material platforms that could cap silicon-photonics share