Taiwan Semiconductor — 8.95/10 — $436.96

Taiwan Semiconductor Manufacturing is a pure-play semiconductor foundry — it does not sell end products, it sells leading-edge wafer fabrication capacity. It commands ~72% of the entire foundry market and ~90%+ of leading-edge (≤7nm) capacity, a position that is strengthening rather than eroding as Samsung foundry (~7% share) struggles and Intel Foundry (<5%) remains years from external leading-edge relevance. No customer can re-source within 12 months, and TSM is a price setter — it pushed through 5–10% advanced-node wafer hikes for 2026 with a further +15% on 3nm.

The AI/HPC megatrend has supercharged the financial profile. High-Performance Computing is now 61% of revenue (up from 46% in early 2024), and every major AI chip — from NVIDIA, AMD, Broadcom, Google, Amazon, and Apple — flows through TSM fabs. Corporate revenue grew ~40% YoY in Q1'26, gross margin expanded to 66.2% (from a 53.1% trough eight quarters earlier), operating margin reached 58.1%, and free cash flow set a record NT$1,002.6B in FY2025 — positive and growing every year, on a net-cash balance sheet (net debt/EBITDA −0.68x).

The one thing keeping the composite out of the high-8s is inverted investor sentiment: this is a crowded, fully-appreciated long where the street already models the bull case (consensus 2026 growth at or above management's own guide), leaving no management-vs-street divergence to exploit. The heavily-weighted, irreducible geopolitical / Taiwan-concentration overhang is the secondary drag on an otherwise pristine quality profile.

Quality gate: PASS (0 of 3 NOs — no cap). All three hard-gate criteria are met: (1) oligopoly — ~72% foundry / ~90%+ leading-edge share, one of ≤3 players controlling >70% of the market; (2) positive & growing FCF — FY2023 NT$292.2B → FY2024 NT$870.2B → FY2025 NT$1,002.6B; (3) 3+ year management track record — 12/12 hit rate, no C-suite turnover, zero red flags.

TSM scores 8.95/10 on this deep-dive screener — a textbook "leader remains leader" quality compounder. The business is a near-monopoly at the leading edge (~72% foundry, ~90%+ advanced-node) positioned at the exact center of the AI/HPC megatrend, and its financial profile is extraordinary: gross margin expanded from 53% to 66% in eight quarters, operating margin from 42% to 58%, and free cash flow grows every year while the share count stays flat and the balance sheet holds net cash. Management under C.C. Wei has delivered a 12/12 hit rate on forward commitments and repeatedly raised long-term CAGR and gross-margin floors — a best-in-class execution machine.

The central tension is that this is a consensus long with no variant perception left. On the inverted sentiment dimension (which asks whether management is bullish about something the street disbelieves), TSM scores just 3/10: the street already models the bull case, all ratings are Buy, targets sit above the current price and are still rising, and the sharpest pushback argues management is being too conservative. There is no management-vs-street divergence to exploit.

The key question for investors: at ~27–29x forward earnings — a ~20% discount to the semi peer median despite ~30% USD revenue growth — is TSM cheap enough to own the AI compute build-out directly, given a genuine, unhedgeable geopolitical tail? The quality bar is cleared with zero caps; the debate is entirely about whether the crowded positioning and Taiwan-concentration overhang are adequately compensated by the discount.

• N2 (2nm) ramp through 2026: entered high-volume manufacturing in Q4'25 at Hsinchu and Kaohsiung with good yield; "fast ramp" in 2026, N2P and A16 (super-power rail) volume in 2H26.
• AI-accelerator CAGR raised to mid-to-high 50s% for 2024–2029; overall corporate revenue CAGR guided to ~25% USD over five years. Watch whether the street revises toward (or past) this.
• Capex step-up to $52–56B in 2026 (from $40.9B) with depreciation rising high-teens% — monitor whether FCF still grows through the heavier investment year.
• Pricing power: ~5–10% advanced-node wafer hikes notified June 2026 (+15% on 3nm) — a second consecutive ~20% ASP year.
• Geopolitical / regulatory: cross-strait tail risk, US Section 232 semiconductor tariffs, export-control tightening, and the live ITC 337-TA-1443 proceeding (preliminary ~late-June 2026, final ~Oct 2026).
• Arizona: Fab 2 tool-in 2026, HVM pulled forward to 2H27; ~30% of N2-and-below eventually US-based — partial mitigation of concentration risk.

For the full financial tables, thematic build-out, and valuation detail, see the Financials, Thematics, and Valuation pages.

Core long for AI-infrastructure and quality-compounder portfolios. The forward P/E (~27–29x) sits ~20% below the semiconductor peer median despite ~30% USD revenue growth, 66% gross margins, and a net-cash balance sheet — the market is pricing in some combination of geopolitical discount and deceleration risk. The primary position-sizing consideration is the binary, unhedgeable Taiwan-concentration risk, which warrants a permanent valuation discount and active position management rather than a maximum-conviction weight. China demand exposure has already reset to ~9% (down from ~22%) on export controls, so the residual China risk is retaliatory/supply-chain rather than demand. The inverted-sentiment score is a reminder that the easy money is priced in — the edge here is quality and durability, not variant perception.