ASML 3Q25 - AI vs. China cliff. What will happen next? Bull case $1,350. Bear case $650.

Cheat Sheets

October 15, 2025

ASML Investment Report - Q3 2025 €7.5B Revenue, 51.6% Margins, AI Boom Continues but China Warning

ASML HOLDING N.V. (ASML)

🔬 Makes Machines That Make Chips | Rev €7.5B, Bookings €5.4B, Margins 51.6%

Q3 2025: AI Demand Strong But China Warning + Vague 2026 Guidance

💰 Worth: $382B | 🏢 Staff: 44,027 | 🌍 Backlog: €33B
👨‍💼 CEO: Christophe Fouquet | 🎯 Makes EUV Machines | 💎 Margins: 51.6%
$999.80
📈 +$16.62 (+1.69%) Today
+73% YTD
Price Target Scenarios

18-Month Horizon (Mid-2027 Targets) - Based on Current Price $999.80

$1,350.00
Bull Case (+35%)
AI Explosion + High-NA
China Stabilizes
📊 Valuation:
2027E EPS: €42 × 45x P/E = $1,350/share
🚀 What's Needed:
AI spending explodes • High-NA sales €5-7B/year • China stabilizes at 15-20% of sales • Logic & memory both boom • 80+ EUV machines per year • Service revenue €8-9B • Margins hit 54-56% • 2027 revenue €42-45B • Market values monopoly at 40-45x earnings
$1,100.00
Base Case (+10%)
Steady AI Growth
China Manageable
📊 Valuation:
2027E EPS: €36 × 42x P/E = $1,100/share
⚖️ What's Needed:
AI grows steadily but not crazy • High-NA revenue €3-4B/year • China drops but stays at 10-12% • 65-70 EUV machines per year • Service revenue €7-8B • 2027 revenue €38-40B • Margins 52-53% • Logic stays strong, memory OK • Market keeps 35-40x valuation
$650.00
Bear Case (-35%)
China Collapse + AI Pause
Chip Cycle Downturn
📊 Valuation:
2027E EPS: €22 × 26x P/E = $650/share
🔻 What Could Go Wrong:
China export rules tighten • China sales fall to 5-8% • AI spending pauses • Chip cycle downturn • Only 45-50 EUV machines sold • High-NA gets delayed • Logic spending cuts • Memory oversupply • 2027 revenue only €28-32B • Margins drop to 48-50% • Market cuts valuation to 25-28x • Tariffs hit equipment • Geopolitics freeze
Bottom Line for Retail Investors
🏆
Beat Expectations, Strong Q4
Revenue €7.5B (hit target) • Profit €2.1B (beat by a bit) • Bookings €5.4B beat estimates • Q4 guidance €9.2-9.8B is strong • Full year +15% growth confirmed • 2026 will be "at least flat" (vague on purpose)
💎
AI Story But It Takes Time
More customers buying (not just TSMC) = good for long-term. Tariffs got sorted which helped. But CEO said AI benefits "only partially hit 2026" - most comes later. €1.3B Mistral AI deal. €33B backlog but chunks are for 2027+. Translation: 2027+ story more than 2026.
📈
Service Revenue Growing
Installed base hit €2B in Q3, €2.1B expected Q4. Service business tied to all the EUV machines out there "developing quite nicely." €1.60 dividend per quarter. Bought back €148M in stock. Margins solid at 51-53%.
⚠️
China 42% Is a Problem
China jumped from 27% to 42% of Q3 sales. Been working through huge backlog. "Significant decline" coming in 2026 but won't say how much. Normal is ~20%. That's a big gap. Real guidance comes January. Geopolitics = wildcard.
📅 Mark Your Calendar: Key Dates & Catalysts
📊 Jan 22, 2026 - Q4 Earnings
THE BIG ONE. Real China guidance finally revealed. Watch: China % for 2026 (high teens = relief, low teens = problem), total revenue guide (need €38B+), margin guidance.
📉 Apr 16, 2026 - Q1 Results
First real proof of China decline in actual numbers. Watch: Did China drop to 25%? 20%? 15%? Are AI/logic orders accelerating to compensate?
🚀 Q4 2026 - High-NA Orders
Next wave of €350M High-NA orders expected. Watch: Who's ordering (Intel/Samsung/TSMC)? How many machines (10+ = serious, 2-3 = just testing)?
⚠️ Wild Cards (Ongoing)
Monitor: US/Netherlands export control updates, Taiwan tensions, China retaliation, major customer delays. Any of these can move stock ±10% instantly.
🔔 Set Alerts: "ASML earnings" • "China semiconductor export controls" • "ASML High-NA orders"

📊 Q3 2025: Beat Estimates But China Just Jumped to 42% of Sales

Revenue €7.5B Hit Target • Profit €2.1B Beat • Bookings €5.4B Beat • Q4 Guide €9.2-9.8B Strong • FY25 +15% On Track • But China Up From 27% to 42% • €33B Backlog Has Big 2027+ Chunk 💪

💰
€7.5B
Q3 2025 Revenue
Inline w/ Guide
⚡
€5.4B
Net Bookings
€3.6B EUV!
🎯
€5.49
EPS Q3
Beat Est.
🚀
51.6%
Gross Margin
Q4: 51-53%

📞 What Management Said on the Call

🎊
AI Boom = More Customers
CEO: AI demand spreading beyond just TSMC. Samsung, Intel, SK Hynix all buying. This matters because it means chip supply won't be limited by one customer. Mistral AI deal: €1.3B for 11% stake to embed AI in their tools. Bookings €5.4B beat expectations. €33B backlog = record high.
🚨
China Problem: 42% in Q3!
CFO admitted China was 42% of sales (up from 27% last quarter). They've been working through a huge backlog. Now expect "significant decline" in 2026 but won't say how much. Translation: If China drops from 42% to normal 15-20%, that's a massive revenue hole.
💎
High-NA: 12-18 Months Away
Their newest €350M machines ran 300K+ test wafers. Should hit full maturity in 12-18 months. Big advantage: Uses proven parts from older machines (no new issues to fix). SK Hynix taking delivery for memory chips. But high-volume sales won't happen until 2028+.
🔬
2026 Guidance: Vague on Purpose
Only said: "2026 won't be below 2025." That's it. Translation: flat to maybe +5%. Market wanted +10-15%. Why so vague? China falling + AI rising = competing forces. Real guidance comes in January. Also: High-NA still loses money "even by 2030" until volumes ramp.

📍 CEO's Vibe: Optimistic about AI long-term. Tariff clarity helped customers make plans. But super blunt about China: jumped from 27% to 42% of sales in one quarter. CFO: "Could it change? Sure." But their honest take is it's going down big in 2026.

🚨 CHINA RISK IS REAL: CFO said they've been working through a massive backlog. China sales are "very high compared to normal levels." Normal is ~20% of revenue. If it drops from 42% back to 20%, that's a €5-7B revenue hole to fill. AI growth better be real to make up for it.

💰 Margin Puzzle for 2026: Less China sales (high margin) = bad. More EUV (high margin) = good. High-NA (losing money) = bad until 2030+. Service revenue mix also matters. Net result for 2026? They wouldn't say. Will tell us in January.

🏦 Backlog Math Doesn't Work: €33B backlog sounds huge. But CFO admitted "healthy chunk" is for 2027 and beyond. Lots of High-NA orders that ship in 2028+. Plus stuff gets moved around. Can't just divide backlog by 12 months and assume that's 2026.

💸 Still Returning Cash: €1.60 dividend per quarter. Bought back €148M in Q3. Total buyback €5.9B so far. Won't finish the €12B program on time. New program coming January 2026.

🔬 R&D Staying High: CFO said they have a "big roadmap ahead" to keep improving machines. But they'll be "diligent" about costs and keep R&D increases "controlled." Translation: spending stays high but won't explode.

💬 Key Things Analysts Asked About

🏦 The China Question: Analysts pushed for specifics. Management wouldn't give numbers. CFO explained they've been working through a huge backlog after not selling enough to China for years. Now it's 42% of sales vs normal ~20%. "Very high compared to normal" he said. Will it change? "Absolutely" - but their honest guess is a big drop next year.

💸 That €33B Backlog Isn't All 2026: CFO confirmed a "healthy chunk" is already for 2027+. Plus lots of High-NA orders that won't ship until 2028. Also stuff gets pushed around. Bottom line: Don't do simple math and assume backlog = next year's revenue. It doesn't work that way.

📊 High-NA Makes Money...Eventually: Right now it's "very low positive margins" (barely profitable). Won't get good until volumes ramp. CFO said even by 2030 it'll still drag down overall margins. Needs lots of machines sold to make the economics work. Next big batch of orders expected late 2026.

🎯 Memory Chip Upgrade is Good News: Analysts worried new DRAM designs would need fewer ASML machines. CEO shut that down: "We do not expect layers to drop." In fact, layers will "continue to grow." New designs are more complex = more machines needed. Not less.

🤖 AI to Orders = ???: CEO admitted: "We wish we had a formula" to predict when AI hype turns into actual orders. The good news only "partially affects 2026" - most benefit comes later. Haven't seen orders really accelerate yet. Bookings were "decent" not "spectacular."

🏭 Can They Make Enough?: CEO said they learned from 2022 chip shortage. Built extra factory space (the slow stuff). For actual machines, they have flexibility to ramp faster. Customers give them "timely heads up" and are "transparent" about needs. Everyone's doing their best to avoid bottlenecks.

🚀 Why ASML Could Keep Going Up
💎
They're the Only Game in Town
Zero competitors make EUV machines. If you want to make 3nm or 2nm chips, you MUST buy from ASML. Each machine costs €350M+. Can't switch to anyone else because no one else exists. Machines last 30 years = service revenue forever.
⚡
AI Needs Advanced Chips
Every AI data center needs cutting-edge chips. More customers (not just TSMC) now buying = supply won't be bottlenecked. AI economics justify paying up for expensive advanced chips. 2nm ramp is proof. Multi-year growth story if AI keeps growing.
👑
Memory Chips Need More Not Less
Big worry was new memory designs would need fewer ASML machines. CEO killed that fear: "Layers will continue to grow, not drop." New designs are more complex = more machines needed. SK Hynix buying High-NA for memory. Good news for years to come.
🚀
Next-Gen Machines Coming Fast
High-NA machines already ran 300K test wafers. Should be fully ready in 12-18 months. No major technical problems (uses proven parts). Orders coming late 2026, big shipments 2028+. Also launched packaging tool with 4x better productivity. New revenue streams opening.
🐻 Why ASML Could Drop
📉
China Could Drop 50%+
CFO admitted China jumped to 42% of Q3 sales. Normal is ~20%. If it drops back to normal that's a massive revenue hit. US export controls getting tighter. Zero specifics on 2026 numbers = bad sign. CFO's honest take: big decline coming.
🏥
That Backlog Is Misleading
€33B sounds great until you realize a "healthy chunk" is for 2027+. Lots of High-NA orders ship in 2028. Stuff gets pushed around. Can't assume backlog = next year's revenue. Management's weak "2026 won't be below 2025" guidance tells you everything.
💸
New Machines Lose Money Forever
CFO was blunt: High-NA is "very low positive margins" now. Needs volume to improve. But "even by 2030" it'll still drag down margins. Less China (high margin) also hurts. If margins don't expand but valuation assumes they will = problem.
⚠️
AI Hype ≠ Orders Yet
CEO admitted: "We wish we had a formula" to predict when AI news becomes actual orders. Recent bookings were "decent" not "spectacular." AI benefits "only partially affect 2026" - most is later. If AI spending pauses or questions arise = order delays. Stock up 73% already pricing in perfection.

This analysis is for informational purposes only and should not be considered investment advice. Please consult with a financial advisor before making investment decisions.