- Equity Research Cheat Sheets
- Posts
- TSM 3Q25 - AI numbers still going up! Bull case $450 Bear case $200
TSM 3Q25 - AI numbers still going up! Bull case $450 Bear case $200
Cheat Sheets
TAIWAN SEMICONDUCTOR MANUFACTURING (TSM)
🔬 Chip Foundry Leader | EPS $2.92 Crushes $2.60 Est (+12%), Net Profit +39%, Revenue +30% to $33.1B
Q3 2025: AI Demand Accelerating | Gross Margin 59.5% | 3nm/5nm Revenue 60% | 2nm Coming 2026 | Guidance Raised
💰 Market Cap: $1.58T | 🏢 Employees: 83,825 | 🌐 70% Foundry Market Share | 74% Revenue <7nm
👨💼 CEO C.C. Wei | 🎯 3nm, 5nm, 2nm Technology | 💎 38 Years Old (Founded 1987) | 🇹🇼 Hsinchu, Taiwan
$298.86
📉 -$5.85 (-1.92%) Today
+123% YTD
Price Target Scenarios
18-Month Horizon (Mid-2027 Targets) - Based on Current Price $298.86
$450.00
Bull Case (+51%)
AI Boom Sustains
2nm Leadership Dominates
📊 Valuation:
2027E EPS: $18.00 × 25x P/E = $450/share
🚀 What's Needed:
AI demand exceeds expectations • Revenue growth accelerates to 35%+ annually • 2nm ramp faster than expected • Apple, Nvidia, AMD all expand volumes • China geopolitical risks fade • Gross margins expand to 62%+ sustained • 3nm/5nm/2nm premium pricing sustained • Arizona/Japan fabs profitable faster • CoWoS advanced packaging capacity doubles • N2P enhanced 2nm captures majority share • Samsung/Intel foundry struggles continue • Hyperscalers increase chip orders 50%+ • Market rerates to 24-26x P/E
$360.00
Base Case (+20%)
Steady AI Growth
Technology Leadership Holds
📊 Valuation:
2027E EPS: $15.50 × 23x P/E = $360/share
⚖️ What's Needed:
AI demand grows mid-40s% as guided • Revenue growth 25-30% annually through 2027 • 2nm ramp meets expectations H1 2026 • Gross margins stabilize 58-60% range • Advanced node (7nm and below) stays 72-76% of revenue • CoWoS capacity supports demand • Arizona fab dilution limited to 2-3% • Smartphone segment stabilizes 25-30% revenue • HPC (AI) remains 55-60% of revenue • China tensions manageable, no major disruptions • Automotive & IoT segments grow steadily • CapEx $40-45B sustains technology leadership • Market values at 22-24x P/E (historical premium range)
$200.00
Bear Case (-33%)
AI Bubble Bursts
Competition Intensifies
📊 Valuation:
2027E EPS: $10.00 × 20x P/E = $200/share
🔻 What Could Go Wrong:
AI spending slowdown or bubble burst • Revenue growth decelerates below 15% • 2nm yields below expectations, delayed adoption • Samsung/Intel foundry catches up • China invades Taiwan or major geopolitical crisis • US export controls tighten dramatically • Gross margins compress to 52-55% from overseas fab dilution • Customer concentration risk (Apple reduces orders) • CoWoS bottleneck persists longer than expected • Tariffs significantly impact margins • Smartphone demand crater in recession • Major customer loss to competitors • CapEx cuts signal weakening demand • Arizona/Japan fabs 5%+ margin dilution • Market derates to 18-20x P/E on growth concerns
Bottom Line for Retail Investors
🏆
Record Smashing Quarter - AI Demand Accelerating
EPS $2.92 CRUSHED $2.60 est (+12%!). Revenue $33.1B beat $32.1B est. Net profit +39% YoY to NT$452B = RECORD. Gross margin 59.5% (+170bps YoY). Operating margin 50% (+240bps YoY). C.C. Wei: "AI market developments continue to be very positive. Increasing AI adoption = more compute demand. Our conviction in the AI megatrend is STRENGTHENING." CFO: "Business supported by strong demand for our leading-edge process technologies."
💎
Raised 2025 Guidance - Mid-30% Revenue Growth!
RAISED 2025 revenue guidance to MID-30% growth (from ~30%). CapEx raised to $40-42B (from $38-42B floor) = more conviction. Q4 guide: Rev $32.8-33.6B, GM 59-61%, OM 49-51%. AI accelerator revenue growing "SLIGHTLY BETTER" than mid-40s% previously announced. CEO: "Very strong signals directly from customers' customers requesting increased capacity." Q3 GM beat by 200bps due to better FX + "better than expected cost improvement efforts." 2026 looks "healthy" per CFO.
📈
Technology Moat Widening - 2nm On Track
Advanced tech (<7nm) = 74% of revenue. 3nm = 23%, 5nm = 37%, 7nm = 14%. HPC (AI) = 57% revenue. CEO: "N2 on track for volume production LATER THIS QUARTER. Good yield = faster 2026 ramp from smartphone + AI." N2P (enhanced 2nm) coming H2 2026 with "further performance and power benefits." A16 chip (best-in-class power delivery) also H2 2026. CFO: "N2 structural profitability is BETTER than N3." CoWoS packaging ~10% revenue now. Apple/Nvidia/AMD locked in for 2nm. Samsung stuck at 3nm yields. Intel years behind.
⚠️
Geopolitical + Tariff Risks + Overseas Fab Dilution
CEO: "Uncertainties and risks from potential tariff impact, especially consumer-related segments. We'll remain mindful and prudent in 2026 planning while investing for future megatrend." Taiwan-China tensions = perpetual sword of Damocles. Arizona fab still 2-3% GM dilution (could widen to 3-4%). Q4 revenue DOWN 1% sequentially despite AI hype. Smartphone only 30% revenue (down from 40%+ peaks). Stock +123% YTD = sky-high expectations. Any AI miss = BRUTAL selloff risk.
📊 What Just Happened: Net Profit +39%, Revenue +30% YoY, EPS Crushes by 11%
Revenue $33.1B (Beat by $1B) • EPS $2.92 (Beat by $0.32!) • Net Income +39.1% to NT$452.3B • Gross Margin 59.5% (+170bps YoY, +90bps QoQ) • Operating Margin 50.6% (+240bps YoY, +100bps QoQ) • 3nm Revenue 23% • 5nm Revenue 37% • HPC (AI) Revenue 57% (Flat QoQ) • Smartphone Revenue 30% (+19% QoQ) • IoT 5% (+20% QoQ) • Automotive 5% (+18% QoQ) • Advanced Tech (<7nm) 74% Revenue • NT$5 Dividend Declared (Q2 2025) • $90B Cash & Securities • 2025 Revenue Growth Raised to Mid-30% from ~30% • CapEx $40-42B for 2025 (Narrowed from $38-42B) • Q4 Revenue Guided $32.8-33.6B (Midpoint $33.2B = -1% Sequential) • Q4 Gross Margin Guided 59-61% (Midpoint 60%) • Q4 Operating Margin Guided 49-51% (Midpoint 50%) 💪
💰
$33.1B
Q3 2025 Revenue
+30.3% YoY
⚡
59.5%
Gross Margin
+90bps Sequential
🎯
$2.92
EPS Q3
Beat $2.60 Est
🚀
+39.1%
Net Profit Growth
NT$452.3B
📞 Key Q3 2025 Earnings Call Highlights
🎊
AI Demand "More Strong Than We Thought"
C.C. Wei: "AI demand actually continues to be very strong, more strong than we thought three months ago. We are getting very strong signals directly from customers' customers, they're requesting increased capacity to support their business needs." Previously guided AI accelerator revenue growth mid-40s%, now "slightly better." On token growth: "The number of tokens increase, it's exponential. I believe that almost every three months, it will be exponentially increased." HPC segment = 57% revenue (flat QoQ but up from H1). Conviction in AI megatrend strengthening.
🚨
3nm & 5nm Crushing = 60% of Revenue!
3nm contributed 23% wafer revenue (iPhone 16 A18 Pro + AI chips). 5nm contributed massive 37% wafer revenue (AI GPUs + smartphone). Combined 3nm+5nm = 60% of total revenue! 7nm another 14%. Total advanced tech (7nm and below) = 74% revenue = pricing power. Analyst: "TSMC's robust earnings are a direct reflection of strong traction at 3nm as well as high utilization at 4/5nm." C.C. Wei: "We continue to observe robust AI-related demand throughout 2025, while non-AI end market has patterned out and is seeing mild recovery." CoWoS advanced packaging approaching 10% of total revenue now.
💎
2nm Coming 2026 = Next Growth Driver
C.C. Wei: "N2 is well on track for volume production later this quarter. With good yield, we expect a faster ramp in 2026, fueled by both smartphone and HPC AI applications." Introduced N2P as extension with "further performance and power benefits on top of N2, volume production scheduled for second half of 2026." Also introduced A16 featuring "best-in-class superpower rail (SPR), best suited for specific HPC product with complex signal route and dense power delivery networks." A16 volume production H2 2026. "We believe N2, N2P, A16, and its derivatives will propel our N2 family to be another large and long-lasting node for TSMC." Wendell: "N2 structural profitability is better than N3."
🔬
Foundry 2.0 = System-Level Optimization
C.C. Wei: "Foundry 2.0 focuses on system performance, not just single-chip technology. One of my customers said that the system performance is very important these days, not only a single chip. Previously, we only looked at the front-end portion. Now it's the whole thing, the front-end, the back-end." Advanced packaging revenue approaching 10% of total revenue = "significant and important for our customer." CoWoS (Chip-on-Wafer-on-Substrate) critical for AI chips. Also building two advanced packaging fabs in Arizona + partnering with OSAT companies. "We can marry in the U.S." to support customers. Heterogeneous integration = moat widening.
📍 CEO on Growth: "Supported by our strong technology differentiation and broad customer base, we now expect our full-year 2025 revenue to increase by close to mid-30% year-over-year in U.S. dollar terms."
🌍 Tariff Concerns: "We understand there are uncertainties and risks from potential tariff impact, especially in consumer-related and price-sensitive segments. We will remain mindful and be prudent in business planning going into 2026 while continuing to invest for the future megatrend."
💰 CapEx Philosophy: "At TSMC, a higher level of capital expenditures is always correlated with higher growth opportunities in the following years. Even as we invest for future growth with this higher level of CapEx spending in 2025, we remain committed to delivering profitable growth."
🔮 Overseas Fabs & Arizona Expansion: "Gross margin dilution from overseas fabs now expected 1-2% for full year 2025 vs 2-3% previously. Thanks to company's overall larger scale." C.C. Wei: "We are preparing to upgrade our technologies faster to end-to-end and more advanced process technologies in Arizona, given the strong AI-related demand. Furthermore, we are close to securing a second large piece of land nearby to support our current expansion plans and provide more flexibility in response to very strong multi-year AI-related demand. Our plan will enable TSMC to scale up to an independent gigafab cluster in Arizona to support needs of our leading-edge customers in smartphone, AI, and HPC applications." Also building two advanced packaging fabs in Arizona plus partnering with OSAT company there.
🚀 Why TSM Bulls Are Right
💎
AI Demand "More Strong Than We Thought 3 Months Ago"
C.C. Wei on capacity planning discipline: "We have more than 500 different customers across all end market segments. As process technology complexity increases, engagement lead time with customers is now at least two to three years in advance. Therefore, we probably get the deepest and widest loop possible in the industry." What's different now vs past cycles: "Right now we pay a lot of attention to our customers' customers. We talk to and discuss with them and look at their applications, be it in the search engine or in social media applications. We talk with them and see how they view the AI application to those functions. Then we make a judgment about what the AI is going to grow. This is quite different as compared with before. We only talked to our customers and had an internal study."
⚡
Technology Moat = Unassailable 2+ Years Ahead
74% of revenue from advanced nodes (<7nm). 3nm ramping fast (23% revenue). 5nm massive (37% revenue). 2nm launching Q4 2025 (per transcript: "later this quarter") with faster 2026 ramp expected. Samsung stuck at 3nm with yield issues. Intel foundry struggling, years behind. China foundries blocked by export controls from advanced equipment. C.C. Wei: "Our technology continues to improve, our customer moving from one node to the next node so that they can handle much more tokens' number in their basic fundamental calculation." Technology node migration + customer design improvements = exponential token handling. TSMC CoWoS packaging = critical bottleneck that competitors can't replicate. Pricing power = gross margins sustained 58-62%.
👑
Customer Concentration = Feature Not Bug
Bears worry: "Apple = 20% revenue, what if they cut orders?" Bull answer: Apple iPhone 16 ramping strong, iPhone 17 coming 2026 with 2nm, Apple designing custom AI silicon = all TSMC. Nvidia: AI chip demand insane, Blackwell ramping, can't get enough capacity, locked into TSMC. AMD: MI300/MI400 AI accelerators, all TSMC. Broadcom: custom AI ASICs for hyperscalers, all TSMC. These customers CAN'T move elsewhere = no alternative foundry with technology + capacity + packaging. Customer concentration = volume certainty + pricing power when you're the only game in town. On China export controls: C.C. Wei: "I have confidence in my customers. Both GPU or ASIC. They all perform very well. If China market is not available, I still think AI's growth will be very dramatic. I have confidence our customers' performance and they will continue to grow, and we will support them." Even with limited China opportunity, still confident in 40%+ AI CAGR.
🚀
Margin Expansion Story Just Beginning
Wendell: "Every year we spend CAPEX based on business opportunity in following few years. If we do our job right, growth of revenue should outpace growth of CAPEX. That is what we've been delivering." "A higher level of CAPEX is always correlated with higher growth opportunity. Next year looks healthy and we are confident on the megatrend." Gross margin 59.5% Q3. Operating margin 50.6%. CFO: "Gross margin dilution from overseas fabs better than expected - now 1-2% vs 2-3% guided for 2025." N3 to reach corporate average margins sometime in 2026. N2 "structural profitability is better than N3." CoWoS packaging ~10% revenue and high-margin. Already generating $90B cash. Analysts raising: Susquehanna $400, Morgan Stanley bullish.
🐻 Why TSM Bears Have Valid Concerns
📉
Stock Up 127% YTD = High Expectations Baked In
TSM traded $134 in October 2024. Hit $307 high = +129%. Now $299 after -1.92% pullback today = +123% YTD still. Stock more than doubled in one year. P/E ratio 33x forward = expensive vs historical 20-25x. ANY disappointment in AI demand, ANY guidance miss, ANY margin compression = stock could crater 20-30% instantly. Q4 guidance actually implies -1% sequential revenue decline (seasonality, but still). If AI spending slows even slightly = stock priced for perfection. Valuation risk extreme. Today's -$5.85 decline shows how fast momentum can reverse. Nvidia down = TSM follows. Macro recession = enterprise IT spending cuts = chip demand plummets. Stock could give back 6-12 months gains in weeks. Counter-argument: When asked about prebuilt/inventory concerns, C.C. Wei: "We don't worry about prebuilt because when you have prebuilt, you have inventory. These days, inventory already goes to very seasonal level and very healthy. So no prebuilt."
🏥
Taiwan Geopolitical Risk = Existential Threat
All of TSMC's most advanced fabs are in Taiwan. If China invades or blockades Taiwan = game over for global tech supply chain. Arizona fab won't be volume production until 2025-2026 and will be 2-3 nodes behind Taiwan initially. US increasingly concerned about Taiwan dependence. Export controls on China could tighten further. Any Taiwan Strait crisis = stock down 40-60% instantly regardless of fundamentals. Warren Buffett sold entire TSMC position in 2023 citing geopolitical concerns. Risk premium should be much higher than current valuation suggests. One military incident = portfolio devastation.
💸
Tariffs + Overseas Fab Dilution = Margin Compression
C.C. Wei: "We understand there are uncertainties and risks from potential impact of tariff policies, especially in consumer-related and price-sensitive market segments." Trump Administration could impose chip tariffs = margin hit. Arizona fab STILL dilutive 2-3% to gross margins, could widen to 3-4% as ramps. Higher US labor costs, construction costs, energy costs = structural margin disadvantage. If forced to move more production to US/Europe for geopolitical/subsidy reasons = margin compression accelerates. Gross margins could fall from 59% to 54-55% = massive EPS hit. Operating leverage reverses.
⚠️
AI Bubble Risk = "What If It's 2000 All Over Again?"
Current AI hype reminiscent of dot-com bubble. Hyperscalers spending $200B+ annually on AI infrastructure with unclear ROI. What if AI adoption slower than expected? What if ChatGPT usage plateaus? What if AI doesn't generate revenue to justify capex? Then hyperscalers cut AI spending 30-50% = TSMC AI chip revenue (now 40%+ of total) collapses. HPC segment falls from 57% revenue to 35% = massive revenue/margin hit. Smartphone segment at 30% revenue Q3 (increased 19% QoQ per transcript but down from 40%+ historical peaks) = limited offset. Samsung catches up on 3nm yields in 2026 = competition. Intel foundry gets act together = share loss. Bear case: TSM revenue growth goes negative, margins compress to 52%, stock trades $150-200.
This analysis is for informational purposes only and should not be considered investment advice. Please consult with a financial advisor before making investment decisions.
