Investment Thesis Summary

Semiconductors are the foundation of every technology trend: AI, EVs, IoT, 5G, and cloud computing. The industry is experiencing a structural supercycle driven by AI demand, with the total semiconductor market projected to reach $1T by 2030. Key sub-segments: AI accelerators (NVIDIA, AMD), leading-edge foundries (TSMC), semiconductor equipment (ASML, Applied Materials), and memory (Samsung, Micron, SK Hynix).

Bull Case

  • AI demand creates a structural supercycle — not a typical cyclical peak but a new growth paradigm
  • Leading-edge manufacturing (3nm, 2nm) is a natural oligopoly: only TSMC, Samsung, Intel can compete
  • CHIPS Act subsidies ($52B US, €43B EU) de-risk manufacturing expansion and ensure secular investment
  • HBM (High Bandwidth Memory) demand growing 5x by 2027, driven entirely by AI workloads
  • Semiconductor equipment companies (ASML, Applied Materials) are picks-and-shovels plays with 50%+ gross margins

Bear Case

  • Semiconductor cycles are notoriously volatile — current AI demand could be pulled forward
  • China developing domestic alternatives could reduce addressable market for Western chipmakers
  • Leading-edge node costs escalating ($20B+ per fab) — ROI increasingly concentrated in top 3 players
  • Memory oversupply cycles return once AI demand normalizes, crushing DRAM/NAND margins
  • Geopolitical risk around Taiwan (90% of advanced chip manufacturing) is existential for the industry

Key Risks

Risk FactorDetailSeverity
Cyclicality Semi industry has 3-4 year boom/bust cycles; AI may not fully decouple High
Taiwan Risk TSMC concentration in Taiwan creates geopolitical fragility for entire tech stack High
China Competition SMIC and domestic players advancing despite export controls Medium
Capex Intensity Leading-edge fabs cost $20B+; ROI timelines measured in decades Medium
Customer Concentration Apple + NVIDIA represent ~25% of TSMC revenue Medium

Competitive Moat Analysis

Technology 95/100

Leading-edge manufacturing requires decades of expertise; only 3 companies can do sub-5nm

Cost Advantage 90/100

Scale economics extreme: $20B fabs amortized over billions of chips

Switching Costs 85/100

Chip design is foundry-specific; switching takes 2-3 years and $100M+

Regulatory Moat 70/100

Export controls create government-backed barriers to entry

Network Effects 50/100

EDA tool ecosystems and IP libraries create vendor lock-in

Key Metrics to Watch

Global Semiconductor Revenue
>$700B
Industry health indicator
AI Chip Revenue Growth
>50% YoY
Validates structural supercycle thesis
TSMC Utilization Rate
>80%
Bellwether for industry demand
HBM Shipments
Doubling annually
AI memory demand indicator
Semi Equipment Bookings
Book-to-bill >1.0
Forward capacity investment indicator
Inventory Days
<90 days
Inventory build signals cycle top

Want to build a custom thesis for any stock or industry?

Build Your Own Thesis →

Explore More Theses

For informational purposes only. Not financial advice. AI-assisted research — always verify data independently. AI Disclaimer

📈 THE FINANCE STACK

Get your weekly market edge. Free.

Market pulse, stock spotlights, and actionable frameworks — delivered every week.

No spam · Unsubscribe anytime · View all issues →