Taiwan Semiconductor — 8.7/10 — $339.04

TSMC is the most strategically important company in the global semiconductor industry and arguably in all of technology. Its foundry monopoly -- commanding ~72% of the global market and 90%+ at advanced nodes -- is strengthening rather than eroding, as Samsung foundry struggles widen the competitive gap and Intel Foundry remains years from relevance. The AI megatrend has supercharged the financial profile: revenue grew 36% in FY2025 to $122B, gross margins expanded to 62% in Q4 2025 (from 53% just 8 quarters earlier), and EPS more than doubled YoY.

HPC/AI now represents 55-60% of revenue, up from 46% in early 2024, with AI accelerator revenue growing at a mid-40s% CAGR that management hints may be revised upward. Every major AI chip -- whether from NVIDIA, AMD, Broadcom, Google, Amazon, or Apple -- flows through TSMC fabs. The CoWoS advanced packaging bottleneck is being addressed with a near-quadrupling of capacity by late 2026, which should unlock further revenue growth. N2 (2nm) volume production began in Q4 2025 with higher structural profitability than N3 and stronger-than-historical customer adoption.

The primary risk is geopolitical -- the possibility of Chinese military action against Taiwan. This tail risk is real, binary, and unhedgeable, warranting a permanent valuation discount. The $165B Arizona investment provides partial mitigation (30% of N2+ capacity) but Taiwan will remain the manufacturing center of gravity. Secondary risks include customer concentration (NVIDIA + Apple = ~36% of revenue), the magnitude of the capex cycle ($41B in FY2025, guided $52-56B in FY2026), and overseas fab margin dilution.

TSM receives a composite score of 8.7/10, reflecting the strongest competitive moat in semiconductors combined with exceptional AI-driven financial momentum and best-in-class management execution -- partially offset by an unhedgeable Taiwan geopolitical tail risk and aggressive capex cycle.

Bull case: TSMC is the sole enabling infrastructure for the AI buildout. Every major AI chip designer -- NVIDIA, AMD, Broadcom, Google, Amazon, Apple -- depends on TSMC fabs. The foundry monopoly is strengthening, not weakening: ~72% global share, ~90%+ at advanced nodes, with Samsung a distant #2 at ~7%. Revenue grew 36% in FY2025 to $122B with gross margins expanding to 62% -- extraordinary for a capital-intensive manufacturer. Management under C.C. Wei operates with Swiss-watch precision: guidance is conservative, execution consistently exceeds expectations, and the full-year 2025 revenue guidance was raised three times. AI accelerator revenue is growing at a mid-40s% CAGR with management hinting it may be revised upward. CoWoS capacity is nearly quadrupling by late 2026, unlocking a key bottleneck. N2 (2nm) volume production began with higher structural profitability than N3.

Bear case: The Taiwan geopolitical risk is real, binary, and existential. A Chinese military action against Taiwan would halt >90% of the world most advanced chip production, with analysts estimating $10 trillion in Year 1 economic impact. Customer concentration is meaningful: NVIDIA (~19%) and Apple (~17%) represent ~36% of revenue. The capex cycle is aggressive ($41B in FY2025, guided $52-56B in FY2026) and Arizona fab dilution of 2-3% on gross margin is a known headwind widening to 3-4% in later years. If AI demand growth disappoints consensus, the return on this capex could compress.

Bottom line: TSMC is a monopoly asset with 20%+ revenue CAGR visibility through 2029, exceptional management execution, and unmatched positioning at the center of the AI megatrend. The geopolitical discount and capex absorption will remain overhangs, but the fundamental quality of the franchise supports a premium multiple.

Key catalysts and monitoring points:

• CoWoS capacity ramp (2026): Target of ~130,000 wafers/month by late 2026 (~4x late 2024 levels). This is the key bottleneck unlock for AI chip production. Watch quarterly capacity additions.
• N2 (2nm) revenue contribution (2026): Volume production started Q4 2025. Revenue contribution could meaningfully exceed the ~10% N3 achieved in its second year given both smartphone and HPC adoption from day one.
• AI accelerator revenue CAGR: Management guided mid-40s% through 2029 but hints it may be higher. Watch quarterly HPC platform mix and AI-specific commentary.
• Arizona Fab 1 performance: Yield already matches Taiwan. Watch for gross margin dilution narrowing as scale economies improve. Premium pricing partially offsets costs.
• FY2026 capex execution ($52-56B guided): Largest capex in company history. Watch for any guidance adjustments and whether revenue growth continues to outpace capex.
• NVIDIA customer concentration: NVIDIA overtook Apple as largest customer at ~19% of revenue. Watch for any AI spending moderation from hyperscalers.
• Taiwan geopolitical developments: Monitor China-Taiwan tensions, US-China relations, and any escalation in military posturing.
• FX sensitivity: Every 1% NT appreciation reduces gross margin by ~40bps. Watch NT/USD movements.

For the full analysis, see the Business Model, Financials, and Valuation pages.

Concerns, Catalysts & Risks -- full analysis

Buy -- TSMC is a foundry monopoly with unmatched positioning at the center of the AI megatrend, exceptional management execution, and 20%+ revenue CAGR visibility through 2029. At $339.04, the stock trades at a premium multiple appropriate for a monopoly asset -- but the premium is justified by the strength and durability of the competitive moat.

The investment case rests on three pillars: (1) The foundry monopoly is strengthening, not weakening -- ~72% global share, ~90%+ at advanced nodes, with no credible challenger emerging. (2) AI demand visibility is unprecedented, with every major AI chip designer dependent on TSMC and management projecting 20%+ revenue CAGR through 2029. (3) Management execution is world-class, with conservative guidance consistently exceeded and capital allocation balancing aggressive growth with healthy FCF.

The geopolitical discount is the key risk management question. The Taiwan risk cannot be diversified away and warrants position sizing discipline. Arizona expansion (30% of N2+ capacity) provides partial mitigation but Taiwan will remain the center of gravity. This is a core holding with appropriate position sizing to reflect the tail risk, not a name to avoid.

What would change the recommendation: (1) Material deterioration in AI demand signals (HPC platform mix declining, AI accelerator revenue CAGR slowing below 30%). (2) Significant geopolitical escalation (military exercises, blockade scenarios). (3) Competitive threat from Samsung or Intel Foundry gaining meaningful advanced node share. (4) Gross margin sustained below 55% due to overseas dilution and pricing pressure. None of these are in the base case today.