Taiwan Semiconductor Manufacturing Company Limited (TSM): what the price assumes
In the published model solve dated 2026-Q2, anchored at $374.36, Taiwan Semiconductor Manufacturing Company Limited (TSM) is priced for today's economics sustained for ~5.8 years. boothcheck publishes no house fair value, target price, or buy/sell rating; individual model outputs and user-controlled scenarios are analytical inputs, not Boothcheck targets. Narrative composed 2026-06-19.
Generated: 2026-07-29 · Exported: 2026-08-01 · Source: https://boothcheck.com/report/TSM
Headline
| Field | Value |
|---|---|
| Ticker | TSM |
| Company | Taiwan Semiconductor Manufacturing Company Limited |
| Sector / Industry | Technology / Semiconductors |
| Current price | $374.36/sh |
| Composition | High Performance Computing 58% / Smartphone 29% / Internet of Things 5% / Automotive 5% / Digital Consumer Electronics 1% / Others 2% |
What The Price Assumes (Inversion)
The assumption today's price embeds, recovered by inverting the valuation.
| Field | Value |
|---|---|
| Inversion basis | whole-company |
| Operating margin (value-band context) | 28.5% |
| Operating margin today | 45.7% |
| Margin compression (value-band) | -17.2pp |
| Must persist for | 5.8y |
| Multiple paid | 47x operating income |
The operating-margin figure is value-band context at year 8: derived from the framework's value band, a separate calculation — not part of the priced-in solve.
Solve inputs: computed at a 11.4% cost of capital; growth searched up to the 46.8% self-funding ceiling; each 1pp moves the implied horizon ~0.7 years.
Reconcile: at the x-ray's 9.3% required return this reads ~39.5%/yr; the models below use their own rates.
How unusual the bet is: extreme
| Reference | Value |
|---|---|
| vs own history | +0.97σ |
| cohort percentile (of 190 peers) | 80 |
| sustained it ~5.8 years at this level | 23% |
| implied end-window share | 2% |
Valuation X-Ray
Every valuation family lands below the price. The price therefore sits beyond what those standard frames encode.
How the valuation models price the stock relative to the market price. Price/FV above 1.0 means the market pays more than that lens defends (expensive); at or below 1.0 the lens can defend the price.
| Family | Median price/FV | Models | Reads |
|---|---|---|---|
| Asset | 4.04x | 5 | expensive |
| Earnings | 5.59x | 4 | expensive |
| Relative | 2.68x | 5 | expensive |
| Growth | 1.72x | 3 | expensive |
Families that call it expensive: Asset, Earnings, Relative, Growth
The models below discount at their own flat-beta convention rates (cost of equity 9.3%, WACC 9.1%); the inversion above states its own rate.
Per-Model Detail (n=17)
| Model | Family | FV | Price/FV | Applicable | Methodology |
|---|---|---|---|---|---|
| DCF Perpetual Growth | Growth | $179.34 | 2.09x | yes | FCF base $29.7B, growth 18% (input: historical growth), terminal g 4.0%, WACC 9.1%, 7yr projection |
| DCF Exit Multiple | Growth | $412.86 | 0.91x | yes | Exit EV/EBITDA: 45.4x / 47.4x / 49.4x (bear / base = today's held flat / bull), 7yr |
| Relative Valuation | Relative | $229.59 | 1.63x | yes | P/E 31.9x (blended: static sector reference 22x + trailing (TTM) 55x), scenarios: 25.8x / 31.9x / 38.0x (bear / base = reference held flat / bull), EV/EBITDA 25.41x |
| Simple DDM | Growth | — | — | no | — |
| Two-Stage DDM | Growth | — | — | no | — |
| Simple Excess Return | Asset | $73.60 | 5.09x | yes | BV/sh $25.17, ROE (TTM) 27.0%, ke 9.3% |
| Two-Stage Excess Return | Asset | $127.74 | 2.93x | yes | 5yr excess ROE then converge to ke=9.3% |
| Discounted Future Market Cap | Growth | $217.99 | 1.72x | yes | Rev $89.1B, growth 18% (input: historical growth; tapered), Terminal P/S: 9.7x / 12.0x / 14.3x (bear / base = today's held flat / bull, cap 12x) |
| Peter Lynch Fair Value | Relative | $139.88 | 2.68x | yes | EPS $6.80, growth 21% (input: historical EPS growth), PEG=2.67 (Overvalued) |
| Margin Trajectory | Growth | — | — | no | — |
| Earnings Power Value | Earnings | $58.57 | 6.39x | yes | Normalized EBIT (5y avg op income, one-time charges added back) $30.10B × (1−18%) / WACC 9.1% → EPV (no growth) |
| Residual Income | Asset | $110.34 | 3.39x | yes | BV $25.17 + 5yr PV of (ROE (TTM) 27.0% − Kₑ 9.3%) × BV; BV grows 8.8%/yr |
| Graham Number | Asset | $62.05 | 6.03x | yes | √(22.5 × EPS $6.80 × BVPS $25.17) — Graham's conservative floor |
| EV/EBITDA Relative | Relative | $130.50 | 2.87x | yes | EBITDA $40.32B × sector EV/EBITDA 16.0x |
| FCF Yield | Earnings | $61.42 | 6.10x | yes | FCF $26537.7M / Kₑ 9.3% — zero-growth perpetuity |
| SBC-Adj FCF Yield | Earnings | — | — | no | — |
| Ben Graham Formula | Earnings | $219.41 | 1.71x | yes | EPS $6.80 × (8.5 + 2×15.0%) × (4.4 / 5.3%) |
| ROIC-Justified P/B | Asset | $92.76 | 4.04x | yes | BV $25.17 × (ROIC 33.6% / WACC 9.1%) |
| P/Sales Sector | Relative | $85.87 | 4.36x | yes | Revenue $89.06B × sector P/S 5.0x |
| PEG Fair Value | Relative | $209.83 | 1.78x | yes | EPS $6.80 × (PEG 1.5 × growth 20.6% (input: historical EPS growth)) → PE 30.9x |
| Earnings Yield | Earnings | $73.51 | 5.09x | yes | EPS $6.80 / required return 9.3% (Rf 4.3% + ERP 5.0%) |
| Funds From Operations Multiple | Relative | — | — | no | — |
| Clinical Phase NPV | Growth | — | — | no | — |
| Merton | Asset | — | — | no | — |
| V5 Mechanical | — | — | — | no | — |
Solvency
| Field | Value |
|---|---|
| Net cash | $64.2b |
| Net debt / NOPAT (after-tax) | -1.93x (net cash) |
| Net debt / operating income (pre-tax) | -1.59x (net cash) |
| Interest coverage | 126.0x |
| Burning cash | no |
Bullet Takeaways
- TSMC manufactures the most advanced logic chips in the world, and the demand signal is unambiguous: first-quarter 2026 revenue reached $35.9 billion, up 40.6% year over year, with gross margin of 66.2% and operating margin of 58.1%, and management guides full-year 2026 revenue to grow above 30% in dollar terms.
- The structural risk is geographic, not financial: the most advanced nodes are produced only in Taiwan, which prohibits overseas manufacturing of its leading-edge chips, so a single island concentrates the supply the entire AI and computing industry depends on.
- Watch the margin path against the build: TSMC plans $52 billion to $56 billion of capital spending in 2026 and expects some gross-margin dilution in the second half from ramping 2-nanometer production and overseas fabs, so the thing to watch is whether pricing power on leading-edge nodes offsets the cost of expanding.
Bull Case
The counterintuitive thing about TSMC is that the headline price multiple is measuring the wrong year. On a trailing full-year basis from its last annual filing, the stock looks priced well past what any standard method supports, with a price-to-earnings ratio near 68. But that snapshot rests on a fiscal year now well behind the company, and the business has grown into a large part of that apparent premium since. In the first quarter of 2026 alone, TSMC reported revenue of $35.9 billion, up 40.6% year over year, with gross margin of 66.2% and operating margin of 58.1%, both above its own guidance, and net income up more than 50% from a year earlier. The obvious narrative says the stock is expensive against the past. The data says the earnings base under the price is materially larger than the past, and growing above 30% this year on management's own outlook. That gap between the stale multiple and the live earnings is the bull's first and best point.
The reason the earnings keep compounding is a manufacturing lead no competitor has closed. TSMC's 3-nanometer node already contributes about a quarter of wafer revenue, and its 2-nanometer node entered high-volume production at the end of 2025 with good yield, which means it is shipping the most advanced chips in the world while rivals are still working to match the prior generation. The peer set makes the position vivid. Intel, the only other company that aspired to leading-edge logic at scale, is running a gross margin of 34.8% and a negative operating margin as its foundry investment cycle weighs on results, while TSMC earns roughly double that gross margin building the same kind of product. The customers tell the same story: the most advanced silicon for the companies driving the AI buildout, including the leaders in AI accelerators, is fabricated at TSMC because there is no real alternative for it. NVIDIA earns a 71.1% gross margin on AI accelerators; the wafers under those accelerators come from TSMC, which is the toll booth on the whole road.
The financial quality underneath is exceptional and easy to overlook amid the growth. TSMC carries net cash on the order of $65 billion, interest coverage of roughly 100 times, and a balance sheet that scores at the top of every solvency measure, while still funding one of the largest capital programs in corporate history out of its own cash flow. It generated free cash flow even in heavy-capex years and returns a growing dividend on top. This is not a speculative bet on a future that might arrive; it is the most profitable, most advanced manufacturer in its industry, compounding earnings at better than 30% with net cash on the balance sheet and self-funded growth. The bull case is that the market is anchoring on a backward-looking multiple while the company prints forward-looking growth, and that a business this dominant deserves to be owned through the noise.
Bear Case
The truth a TSMC holder would rather not dwell on is that the entire investment rests on the stability of one island. The most advanced chips the company makes, its 2-nanometer node, are produced only in Taiwan, and Taiwan law prohibits manufacturing its leading-edge semiconductors abroad. That is not a diversification gap the company can simply close with capital. It is a deliberate national policy, the so-called silicon shield, which means the supply the AI and computing industry depends on most is concentrated at a single point of geographic and geopolitical failure. Every other risk in the story is downstream of this one. A blockade, an earthquake, a coercive action, or even a sustained escalation in cross-strait tension would not dent earnings at the margin; it would threaten the source of the world's most advanced logic. The price does not appear to carry much discount for a tail risk that, if it arrived, would be close to uninsurable.
The Arizona build is the visible response, and it is both real and revealing. TSMC has committed about $165 billion to six fabs and packaging facilities in Arizona, and the first is already producing advanced chips for marquee customers. But the project underscores the problem rather than solving it: the truly leading-edge nodes stay in Taiwan by law, so the overseas capacity is a generation or more behind the frontier, and it comes at a higher cost. Management itself flags gross-margin dilution in the second half of 2026 from the combination of ramping 2-nanometer production and overseas expansion. The reshoring that reassures Washington also drags on the margins that justify the multiple. There is a customer-concentration angle layered on top: North America now contributes around 75% of revenue, so the company is exposed both to a handful of large customers and to the trade and tariff politics between the two governments that matter most to it.
Then there is the cyclical reality the current boom obscures. Semiconductors are cyclical, and demand tied to a single wave, the AI capital-spending cycle, is the kind of demand that looks permanent right up until it normalizes. TSMC's customers in AI accelerators are earning extraordinary margins today, NVIDIA at 71.1% gross margin, which is exactly the condition that pulls in competition and capacity across the industry and eventually compresses the cycle. If AI infrastructure spending slows, or if the hyperscalers digest the capacity they have bought, the orders that drive TSMC's leading-edge utilization soften, and a business running near-record margins on near-record utilization has more room to fall than to rise. The bear case is not that TSMC is a bad company; it is plainly a great one. It is that a great company priced for continued acceleration, sitting on a geographic fault line, with margins set to dilute and a demand cycle near its peak, is carrying risks the price treats as remote.
Valuation
Read on its trailing record, TSMC looks priced beyond what any single method supports. Across the families we use to triangulate, the price sits above asset value, above earnings power, above peer multiples, and above the forward-growth methods. The peer-multiple lens is the closest, putting the price at a bit under twice where a blended industry multiple lands; the asset and earnings-power lenses sit further out. Taken at face value, that pattern says the market is paying a premium no static frame can justify. But the face value here carries an important qualifier that changes the reading: the trailing financials underneath these methods come from the company's last full annual record, and TSMC's reported results since then are materially larger. A trailing price-to-earnings ratio near 68 is dividing today's price by an earnings figure the company has already grown well past. So the honest way to engage these methods is to treat them as the upper bound of the premium, computed against a base the live business has outrun.
Reconciling the two systems is the whole exercise. The static methods say expensive against the past; the company's own 2026 results, with revenue up more than 40% year over year and operating margin above 58%, say the earnings base is climbing fast enough to absorb a real portion of that gap. This is the difference between a stock that is expensive because the market has detached from fundamentals and one that is expensive because the fundamentals are growing into the price. TSMC reads more like the second. The growth-method read is the one to weight, because for a company compounding earnings above 30% with a manufacturing lead on the leading edge, the static lenses structurally understate the value of durable growth. None of this produces a target, and the spread between price and the trailing methods is genuine; the point is that the spread is narrower against the forward earnings than the stale multiple implies.
What the price is ultimately underwriting is durability: that TSMC keeps its process lead, that AI and high-performance-computing demand stays strong, and that the leading-edge margins hold even as 2-nanometer ramps and overseas capacity dilutes them. The balance sheet removes solvency from the list of worries entirely. The company holds net cash on the order of $65 billion, covers interest more than a hundred times over, and funds a $52 billion to $56 billion capital program from its own cash flow while still paying a growing dividend. There is no leverage risk and no funding risk here. The risk is concentrated instead in two places the financial statements cannot price: the geographic concentration of leading-edge production in Taiwan, and the cyclicality of the demand wave the current margins rest on. The buyer at this price is paying a premium for the best manufacturer in the industry and accepting those two exposures in exchange.
Catalysts
The first quarter of 2026 set the tone for the year. TSMC reported revenue of $35.9 billion, up 40.6% year over year and ahead of consensus near $35.5 billion, with gross margin of 66.2% and operating margin of 58.1%, both above the high end of its guidance, and net income up 58.3% year over year. The driver was the same one powering the whole sector: strong demand for advanced nodes from high-performance-computing and AI applications, with the 3-nanometer process contributing about 25% of wafer revenue. Management raised the bar for the year, now expecting full-year 2026 revenue to grow above 30% in dollar terms, and guided second-quarter revenue to a range of $39.0 billion to $40.2 billion, roughly 10% sequential growth.
The technology roadmap is the structural catalyst. TSMC's 2-nanometer node entered high-volume manufacturing in the fourth quarter of 2025 with good yield, which keeps the company a generation ahead of its competitors and protects the pricing power that underpins its margins. The flip side, and the one number to track, is margin: management expects some gross-margin dilution in the second half of 2026 from the cost of ramping 2-nanometer and from overseas expansion, so the question through the year is whether leading-edge pricing offsets that drag. Capital spending is being pushed toward the high end of a $52 billion to $56 billion range for 2026, tilted to advanced nodes, specialty technology, and advanced packaging, a signal of how much capacity the company expects to need.
The geopolitical and reshoring thread is the wildcard. TSMC has committed roughly $165 billion to its Arizona buildout, with a third fab under construction and applications filed for a fourth fab and a first advanced-packaging facility, and the first Arizona fab is already producing advanced chips for large U.S. customers. The expansion locks in U.S. market access and hedges tariff risk, but the most advanced nodes remain in Taiwan by law, so the strategic significance is as much political as financial. The events most likely to move the stock are the quarterly margin trajectory as 2-nanometer ramps, the pace of AI-driven order growth from its largest customers, and any shift in the trade and security relationship between the United States and Taiwan.
Methodology Note
- Priced-in inversion: the valuation is inverted on the current price to recover the operating-income growth, duration, and steady-state margin the price embeds (ROE for financials, FFO growth for REITs).
- Valuation x-ray: the valuation models, grouped into four families (asset, earnings, relative, growth). Each model is expressed as a price/FV ratio (distance from price), not a point fair-value estimate. The spread across families is the disagreement.
- Solvency: net cash/debt, net-debt-to-NOPAT, interest coverage, and share-count CAGR from EDGAR financials (net debt / FFO and fixed-charge coverage for REITs; regulatory-capital framing for financials).
- Peer cohorts: per-segment comparables with deep-linkable SEC filing citations.
Fundamentals sourced from SEC EDGAR filings. Current price from Databento. The priced-in inversion and valuation x-ray are computed by the boothcheck engine; narrative composed by AI from the structured data.
Sources
TSMC Q1 2026 results, April 2026 · TSMC Q1 2026 earnings commentary, April 2026 · TSMC Q1 2026 commentary, April 2026 · AppleInsider and Stimson Center analysis, 2026 · tech-insider.org and AInvest reporting, 2026 · TSMC Q1 2026 results; Manufacturing Dive, April 2026 · TSMC Q1 2026 earnings call; Investing.com, April 2026 · tech-insider.org and Finviz reporting, 2026