GlobalWafers
Capital-intensive upstream materials manufacturer selling silicon wafers under a mix of long-term agreements (LTAs, often with customer prepayments) and spot; cyclical, commodity-plus-specialty economics where scale, yield and 300mm mix drive margin
Earnings, margins, COGS & capex
A cyclical wafer pure-play at the bottom of a demand/pricing trough. FY2025 revenue fell ~3.2% to TWD 60.6B and margins compressed hard (gross ~31.6%->24.1%, operating ~22.5%->14.3%) as wafer volumes and pricing softened while the company absorbed start-up costs and depreciation from new capacity. Net income fell ~26% to TWD 7.31B (EPS 15.29, vs ~21.06 in FY2024), missing consensus (~TWD 7.58B). Management framed Q1 FY2026 as the cyclical low; reported net profit rose ~30.2% YoY but that was driven by a mark-to-market gain on the 13.67% Siltronic stake, not operations (Q1 gross margin 20.8%, operating margin ~10.5%, net margin ~13.6% - the net line flattered by Siltronic). The dividend was cut ~60% to NT$2.00. The forward story is an H2 2026 pricing recovery (especially 12-inch), declining capex, and the ramp of US capacity.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~76¢ is cost of goods and ~10¢ operating expense, leaving ~14¢ of operating profit (~12¢ net).
Revenue trend
Margins
down sharply YoY; guided to improve H2 2026 on pricing
compressed from ~22.5% FY2024 on lower utilization + new-fab start-up costs
Q1 FY2026 distorted upward by Siltronic mark-to-market, not core ops
COGS structure
Dominated by fixed cost of high-capacity crystal-growth and wafering fabs (depreciation of new plant is a rising share), polysilicon feedstock, quartz crucibles, energy, and specialty inputs (e.g. gallium for compound wafers, whose price management says has roughly doubled). High operating leverage means utilization swings drive gross margin more than input cost.
Capex
Peak-cycle capex funded the ~$7.5B multi-year global expansion, headlined by the Sherman, TX 300mm fab (opened May 2025; Phase II production started Feb 2026; up to $400M US CHIPS funding across TX + Missouri) plus capacity in Asia and Europe. Management states capex has peaked and steps down in 2026 as projects move from build to ramp.
Latest earnings
Headline EPS TWD 3.97 (vs 3.05 Q1 FY2025, +30% YoY) beat on a Siltronic mark-to-market gain; underlying operations were at/near the cycle trough (gross margin 20.8%, operating margin ~10.5%, revenue -10.3% YoY) - a low-quality, non-operating beat
FY2026 revenue flat to slightly higher; gross margins improving through the year; capex declining; H2 2026 pricing improvement expected, led by 12-inch. Management frames Q1 as the cyclical bottom with recovery arriving earlier than expected.
- Q1 FY2026 revenue
- TWD 13.98B
- Q1 FY2026 gross margin
- 20.8%
- Q1 FY2026 operating margin
- ~10.5%
- Q1 FY2026 EPS
- TWD 3.97 (+30% YoY, Siltronic-driven)
- Siltronic stake
- 13.67% (mark-to-market drives non-op volatility)
- FY2025 EPS
- TWD 15.29 (net income TWD 7.31B)
Growth drivers
- AI/data-center compute driving demand for advanced 300mm wafers (leading-edge logic + HBM/DRAM) and high utilization on 12-inch lines
- Semiconductor onshoring — first-mover 300mm wafer supply inside the US (Sherman) capturing CHIPS-era supply-security demand and premium LTAs
- H2 2026 wafer price increases, especially 12-inch, as utilization and input costs rise
- Specialty/compound wafers (SOI, FZ, gallium-based, SiC/GaN adjacencies) as a higher-margin mix shift
- Long-term agreements with prepayments that underwrite capacity and smooth the cycle
Bull & bear
A trough-cycle entry into the #3 name in a rational 5-player oligopoly, with capex peaking, an early pricing recovery underway, and a scarce strategic asset - the only 300mm wafer fab on US soil - just as AI compute and onshoring re-rate the whole silicon-materials supply chain.
- Q1 FY2026 called the cycle bottom by management; H2 2026 12-inch price hikes plus rising utilization can drive gross margin back toward the high-20s/30s of prior peaks
- Capex has peaked and steps down into 2026, so free cash flow can inflect from deeply negative toward positive even on flat revenue - a large FCF swing
- Sherman, TX is a genuinely scarce asset: first-mover US 300mm supply with CHIPS support wins supply-security LTAs and pricing power as customers de-risk from Asia
- Oligopoly discipline (5 players ~82% share, brutal entry barriers) means the industry does not chase share with price the way a fragmented commodity would
- Trough earnings understate mid-cycle power: FY2025 EPS of 15.29 is well below normalized, so today's optically high P/E is cyclical, not structural
A capital-destroying two-year stretch (deeply negative FCF, ~26% earnings decline, dividend cut ~60%) in a demand-trough where the optical 'beat' is a Siltronic paper gain, the balance sheet carries net debt from a build-out that ramps into an uncertain pricing recovery, and the stock already discounts a strong upcycle.
- FY2025 was ugly on the fundamentals: revenue -3.2%, gross margin ~31.6%->24.1%, net income -26% to TWD 7.31B (a consensus miss), and the dividend cut ~60% to NT$2.00 - management voting with the payout
- The Q1 FY2026 EPS rise was driven by mark-to-market on the Siltronic stake, not operations; underlying gross margin was 20.8% and operating margin only ~10.5%
- New global + US capacity comes online into a soft market, loading depreciation and start-up costs before utilization justifies it - margin drag if the recovery slips
- Valuation is demanding: ~$18-20B market cap on TWD 7.31B trough net income is a very high trailing multiple (~79-87x) that already prices a robust upcycle; disappointment de-rates it hard (the stock fell ~9.7% on Jul 7 2026)
- Structurally out-scaled by Shin-Etsu and SUMCO at the leading edge, and exposed to AI-capex, FX, gallium/polysilicon input costs, and US-China trade risk
What it is worth
Cyclical reverse-DCF / normalized-earnings sanity check plus peer read-across (vs Shin-Etsu 4063, SUMCO 3436, Siltronic WAF). No 5-tab model; the point is that trailing multiples are near-meaningless at trough earnings.
AI-capex wobble or delayed pricing recovery leaves new capacity under-utilized; margins stay compressed, FCF recovery slips, and the demanding multiple on trough earnings de-rates materially. Note: TPEX-listed Taiwan equity - context only, not a buy/own call; not financial advice.
Gradual 2026-2027 recovery: revenue flat-to-up per guidance, margins recover partway, FCF inflects positive; the stock compounds with the cycle rather than re-rating sharply - fairly valued on mid-cycle earnings.
Fast H2 2026 pricing recovery + Sherman ramp restore gross margin toward high-20s/30s and normalized EPS well above FY2025's 15.29; FCF turns firmly positive as capex rolls off, supporting a re-rate on normalized (not trough) earnings.
At ~TWD 576-640B market cap (~$18-20B USD) on FY2025 net income of TWD 7.31B, the trailing P/E is optically very high (~79-87x; stockanalysis shows ~82x) - but FY2025/Q1 FY2026 are cycle-trough earnings distorted by depreciation from unramped capacity and Siltronic noise. The price implies the market underwrites a mid-cycle recovery: normalized net income well above trough as utilization, 12-inch pricing and the FCF inflection (capex roll-off) play out. The thesis lives or dies on whether the H2 2026 pricing recovery and Sherman ramp arrive on schedule; the ~9.7% drop on Jul 7 2026 shows how sensitively the multiple reacts.
SWOT
Strengths
- World #3 silicon-wafer producer (~15% share) in a 5-player oligopoly (top-5 ~82% combined) with high barriers to entry
- Full product breadth — polished, epi, annealed, diffused, SOI, FZ, and compound wafers across 200mm/300mm - and a deep, sticky customer roster (TSMC, Texas Instruments, and effectively every major foundry/IDM)
- First-mover on 300mm wafer manufacturing inside the US (Sherman, TX) aligned with CHIPS-era supply-chain security demand
- Long-term agreements with customer prepayments underwrite utilization and dampen cyclicality
Weaknesses
- Deeply cyclical, capital-intensive economics — gross margin swung from ~31.6% to 24.1% in one year and FCF has been sharply negative two years running
- Reported earnings polluted by non-operating Siltronic mark-to-market swings, obscuring true operating trajectory
- New US/global capacity adds depreciation and start-up cost ahead of full utilization, a near-term margin drag
- Net-debt position after the build-out — balance-sheet flexibility reduced until capex rolls off and FCF turns; dividend cut ~60% to NT$2.00
Opportunities
- AI-driven 300mm demand and rising utilization enabling H2 2026 price increases
- US onshoring premium: sole domestic 300mm wafer source can command supply-security LTAs and CHIPS support
- Higher-margin specialty/compound (SOI, FZ, gallium, SiC/GaN adjacencies) mix shift
- Cyclical operating leverage - at mid-cycle utilization, normalized earnings materially exceed FY2025 trough
Threats
- Wafer oversupply / pricing pressure if AI demand disappoints or foundry capex pauses
- Larger, better-capitalized Japanese leaders Shin-Etsu and SUMCO out-investing in leading-edge 300mm
- Input-cost inflation (gallium roughly doubled per management; polysilicon, energy) squeezing margin
- FX and geopolitical exposure — Taiwan HQ, global fabs, US/China trade friction, and Siltronic equity-price risk
Moats, dependencies & bottlenecks
Moats
Top-5 makers hold ~82% of a market where a greenfield 300mm fab costs billions and takes years to qualify - near-prohibitive entry barrier.
Wafers are qualified per-customer per-process over long cycles; switching risks yield, so incumbent LTAs are sticky.
Full polished/epi/SOI/FZ/compound range across 200/300mm lets it serve logic, memory, power and analog from one supplier.
Moderate-Emerging Only domestic 300mm wafer source in the US - a supply-security moat backed by CHIPS, still ramping and unproven at scale.
Dependencies
Texas Instruments, Intel, Samsung, GlobalFoundries, Micron) Revenue tracks foundry/IDM capex and utilization; a capex pause hits volumes and pricing directly.
Polysilicon + quartz crucible + specialty input (gallium) suppliers Feedstock cost/availability drives COGS; gallium price has roughly doubled per management.
Up to $400M supports TX/Missouri economics; disbursement and onshoring policy are political variables.
financial holding 13.67% stake creates non-operating mark-to-market P&L swings unrelated to the wafer business.
parent / affiliate SAS is the controlling shareholder; corporate/capital decisions are aligned with the parent group.
Advantages
- Trough-cycle earnings with a clear capex roll-off = large FCF inflection potential
- Scarce strategic asset (US 300mm) with policy tailwind and supply-security pricing power
- Rational oligopoly structure that protects industry pricing discipline
- Broad, qualified customer base across logic, memory, power and analog end-markets
Weaknesses
- Extreme cyclicality and capital intensity with two years of deeply negative FCF
- Low-quality recent 'beat' driven by a financial-asset paper gain, not operations
- ~60% dividend cut (to NT$2.00) signaling near-term cash caution
- Sub-scale versus Shin-Etsu/SUMCO at the leading edge; a price-taker in downturns
Bottlenecks
- Cyclical wafer pricing + utilization - the single biggest margin lever, outside company control
- Ramp/yield qualification of new 300mm capacity (Sherman) before it is margin-accretive
- Balance-sheet capacity: net debt limits flexibility until capex rolls off and FCF turns
- Input-cost inflation (gallium, polysilicon, energy) against LTA-fixed pricing
- Non-operating Siltronic mark-to-market noise obscuring the true operating signal
Top signals & trends
Top signals
The core of the recovery thesis; watch realized ASPs and gross margin trajectory.
Management says capex peaked; the FCF inflection is the key balance-sheet catalyst.
Execution proof point for the US 300mm first-mover premium.
Non-operating noise; strip it out to read core operations.
Leading indicator of wafer demand; an AI-capex wobble is the main downside risk.
Trends
High utilization on 12-inch logic/HBM/DRAM lines underpins the pricing recovery.
Structural demand for US/allied wafer supply; GlobalWafers is the domestic 300mm first-mover.
Legacy/power end-markets weak in 2025-2026, a drag on the mix.
Raises COGS but also underwrites the case for new-contract price hikes.
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.
Polysilicon feedstock supplier.
Polysilicon / semiconductor-grade materials.
Polysilicon feedstock.
Process/inspection equipment for wafer fabs (representative; specialized wafer-shaping tooling from multiple vendors).
Largest foundry customer; leading-edge 300mm logic wafers.
Major analog/embedded IDM customer, named revenue source.
Leading-edge logic and US-onshoring demand.
US specialty foundry customer.
Memory/HBM demand for 300mm wafers.
Power/analog and automotive-chip demand.
Global #1 silicon-wafer maker (~18% share), diversified chemicals giant with the deepest balance sheet and leading-edge 300mm capacity.
Global #2 (~17%), Japanese 300mm-focused pure-play; shifting capacity toward high-end AI-grade 300mm.
German #4-5 wafer maker; GlobalWafers holds a 13.67% stake (a 2020-22 acquisition attempt was blocked on German regulatory clearance). Competitor and financial holding.
Korean wafer maker, private (SK Group subsidiary); strong in 300mm and SiC, backed by SK captive/Korean demand.
Shanghai-listed Chinese 300mm wafer maker (Zing Semiconductor); state-backed, ramping domestic 12-inch capacity to serve China's onshoring push - the rising low-end/mid-tier threat.