(Almost) The Only Game in Town

Main Assertion

TSMC is undervalued. That might be insane for the a top 10 largest company in world by market cap, but a basic considering of the facts and status within the broader economy makes this self-evident:

• TSMC is NOT merely a cyclical semiconductor manufacturer; it is critical infrastructure for the global digital economy.
• Its combination of scale, process leadership, capital discipline, and customer entrenchment supports sustainably high margins and above-GDP growth.

I therefore offer a valuation that considers it's very components and overall valuation holistically, not merely through abstraction. Notably:

• The current valuation embeds mean-reversion assumptions that ignore the durability of advanced-node dominance and AI-driven demand.
• On reasonable long-term assumptions, TSMC supports a meaningfully higher intrinsic value than implied by today’s market multiple.

Argument in Favour of a Stronger Valuation

Especially:

• AI and HPC wafer intensity: Advanced logic demand scales super-linearly with compute complexity, not unit volumes. TSMC's world class R&D ensures success.
• Process gap persistence: Leading-edge nodes (N3, N2 and beyond) reinforce a winner-takes-most cost curve. TSCM is basically the only company in the race.
• Customer lock-in: Fabless leaders design around TSMC’s process roadmap, structurally reducing switching optionality. They're the only game in town!
• Capital efficiency inflection: As node shrink slows, incremental capex increasingly converts into free cash flow rather than transient advantage.

Possible Mitigating Risks

There are a few issues, especially:

• Geopolitical concentration: Self-explanatory. TSMC might not exist, at least in it's current form, if a certain regional bully gets its way.
• Customer bargaining power optics: Large customers appear powerful, but in practice lack viable alternatives at scale.
• Node transition execution: Yield slippage at leading edge would delay margin normalization.
• Regulatory pressure: Export controls may shape mix, but do not negate global compute demand.

Assumptions (Explicit, Not Consensus)

• Revenue growth normalizes below historical peaks but remains structurally elevated.
• Net income margins compress modestly from recent highs but remain far above industry averages due to scale and technological moat.
• Share count remains broadly stable.
• No heroic terminal assumptions: valuation is driven by cash generation, not terminal multiple expansion.

Valuation Framework

Here's where we get to the heart of the matter. Consider the following...

TSMC’s reported trailing data shows:

• Revenue (LTM): NT$3.81T
• Net income margin (LTM): ~45%
• Net income (LTM): NT$1.72T

Rather than applying a market-average semiconductor multiple, it's more sensible to treat TSMC as a platform monopoly with regulated-utility-like inevitability and software-like margins.

Base-case intrinsic framework:

• Sustainable growth above global GDP
• Margins structurally double the industry average
• A terminal multiple reflecting durability, not cyclicality

This supports a future earnings multiple materially above today’s implied forward PE, even after margin normalization.

Conclusion

The market prices TSMC as a cyclical manufacturer with transient pricing power. The financial reality supports pricing it as irreplaceable infrastructure with compounding economics.

Therefore, I argue that it is undervalued relative to actual worth!