TSM
Taiwan Semiconductor Manufacturing Co.
AI Foundry & Connectivity
~$410
Mkt Cap ~$2.1T
52W: $184.62 – $420.00
Next earnings: Q2 2026 (~16 Jul)
↓ Add (pullback)
Entry zone: $340 – $375
The market prices TSMC as an AI proxy with a persistent Taiwan-concentration discount; what it underweights is that the US-Taiwan tariff deal and $465B Arizona buildout are structurally compressing that discount — meaning TSM deserves to re-rate toward a global tech monopoly premium, not trade at a 30% haircut to the semiconductor sector median forward P/E it currently occupies.
TTM P/E
33.6×
Fwd P/E
25.8×
PEG 0.53 (g=49% NTM)
Rev YoY
+40.6%
Q1 2026 USD
Gross Margin
66.2%
Q1 2026
FCF Margin
30.7%
Q1 2026
HPC Rev %
61%
of Q1 2026 rev
Thesis Durability
A
Variant perception: Geopolitical discount persists despite structural de-risking via Arizona buildout and US-Taiwan tariff framework.8
TAM: TSMC guides AI revenue at >mid-40% CAGR through 2029; CC Wei in Q3 2025 noted it's tracking "a little better than that." Full-year 2026 USD revenue guidance upgraded to "above 30%" after Q1 2026 — from "close to 30%" at Q4 2025 earnings — management confirming the upside scenario is materialising.2,3,7 Combined hyperscaler capex commitments (~$600B in 2026 guided) are over-subscribed against TSMC's available leading-edge capacity.3,7
CoWoS capacity monopoly: TSMC is the primary supplier of CoWoS advanced packaging — the critical chokepoint for AI GPU production. Capacity is ramping from ~75–80K wafers/month (end-2025) to ~130K/month (end-2026); current run-rate estimated near ~100K/month. Nvidia has secured ~650K CoWoS wafers for 2026 (+76% YoY per KeyBanc), representing >50% of current annual supply. AMD's allocation is ~80K at TSMC — less than 7%. Nvidia's 2027 allocation projected at ~840K; AMD at ~136K (+70% YoY). CoWoS scarcity is a pricing power dynamic for TSMC: demand is pre-committed before capacity even lands.11
Moat: ~70% global foundry market share at leading-edge. Samsung and Intel Foundry are structurally behind on yield at N3/N2 — TSMC's technology lead at advanced nodes (2nm, A16) is 12–18 months over the nearest competitor. Switching cost is effectively zero for customers but the cost of failure is existential — no chip designer can afford a miss on tape-out.1
Bear case: Advanced fab competitors (Samsung, Intel Foundry) close the yield gap at N2 by 2028. DeepSeek-style inference efficiency permanently reduces AI accelerator demand. China invasion scenario remains a tail risk but the Arizona buildout is incrementally hedging it.
TAM: TSMC guides AI revenue at >mid-40% CAGR through 2029; CC Wei in Q3 2025 noted it's tracking "a little better than that." Full-year 2026 USD revenue guidance upgraded to "above 30%" after Q1 2026 — from "close to 30%" at Q4 2025 earnings — management confirming the upside scenario is materialising.2,3,7 Combined hyperscaler capex commitments (~$600B in 2026 guided) are over-subscribed against TSMC's available leading-edge capacity.3,7
CoWoS capacity monopoly: TSMC is the primary supplier of CoWoS advanced packaging — the critical chokepoint for AI GPU production. Capacity is ramping from ~75–80K wafers/month (end-2025) to ~130K/month (end-2026); current run-rate estimated near ~100K/month. Nvidia has secured ~650K CoWoS wafers for 2026 (+76% YoY per KeyBanc), representing >50% of current annual supply. AMD's allocation is ~80K at TSMC — less than 7%. Nvidia's 2027 allocation projected at ~840K; AMD at ~136K (+70% YoY). CoWoS scarcity is a pricing power dynamic for TSMC: demand is pre-committed before capacity even lands.11
Moat: ~70% global foundry market share at leading-edge. Samsung and Intel Foundry are structurally behind on yield at N3/N2 — TSMC's technology lead at advanced nodes (2nm, A16) is 12–18 months over the nearest competitor. Switching cost is effectively zero for customers but the cost of failure is existential — no chip designer can afford a miss on tape-out.1
Bear case: Advanced fab competitors (Samsung, Intel Foundry) close the yield gap at N2 by 2028. DeepSeek-style inference efficiency permanently reduces AI accelerator demand. China invasion scenario remains a tail risk but the Arizona buildout is incrementally hedging it.
High conviction variant view. The monopoly moat at leading-edge is real, measurable, and widening into N2/A16. The thesis is not fully consensus — the geopolitical discount still exists and is the source of valuation upside.
Business Quality
A
Margin trajectory: Gross margin expanded from 53.5% (Q1 2024) → 58.8% (Q1 2025) → 66.2% (Q1 2026).1,2 Three straight years of margin expansion driven by leading-edge mix shift and pricing power on N2/N3.
Operating margin: 48.5% (Q1 2025) → 58.1% (Q1 2026). Operating leverage is confirming thesis, not narrative.2
FCF: FY2025 operating cash flow NT$2.3T (~$73B), capex $40.9B → FCF ~$32B. FCF margin compressed by deliberate capex ramp. 2026 capex $52–56B (guided toward high end per Q1 2026 call) is demand confirmation, not margin destruction.5
EPS revisions: 8 consecutive quarterly beats. Q1 2026 EPS $3.49 per ADR beat consensus of $3.26 by 7.1%.2
Income quality: Q1 2026 net profit margin 50.5%, up 7.4pp YoY and 2.2pp QoQ. Operating leverage converting revenue growth into bottom-line outperformance at exceptional rates — consistent across three consecutive quarters of margin expansion.11
Cash generation: Q1 2026 operating cash flow $22.1B (61.6% of revenue). Q1 2026 FCF $11.0B (30.7% of revenue) — compressed vs FY2025 (35.1% OCF-to-revenue) due to accelerating capex, but absolute FCF growing. Balance sheet: $105.5B cash and marketable securities vs $31.7B debt — net cash fortress, no financing risk.11
Form 4 / insider activity: TSM is a Taiwanese-listed company; Form 4 filings are not applicable. Yahoo Finance notes First Eagle Investment reduced its stake in Q1 2026.9 No insider cluster signal available.
Customer concentration: Apple ~20% of revenue, NVDA significant and growing. Hyperscaler direct demand now pulling capacity independently of chip designer intermediaries — structurally reduces single-customer risk.
Operating margin: 48.5% (Q1 2025) → 58.1% (Q1 2026). Operating leverage is confirming thesis, not narrative.2
FCF: FY2025 operating cash flow NT$2.3T (~$73B), capex $40.9B → FCF ~$32B. FCF margin compressed by deliberate capex ramp. 2026 capex $52–56B (guided toward high end per Q1 2026 call) is demand confirmation, not margin destruction.5
EPS revisions: 8 consecutive quarterly beats. Q1 2026 EPS $3.49 per ADR beat consensus of $3.26 by 7.1%.2
Income quality: Q1 2026 net profit margin 50.5%, up 7.4pp YoY and 2.2pp QoQ. Operating leverage converting revenue growth into bottom-line outperformance at exceptional rates — consistent across three consecutive quarters of margin expansion.11
Cash generation: Q1 2026 operating cash flow $22.1B (61.6% of revenue). Q1 2026 FCF $11.0B (30.7% of revenue) — compressed vs FY2025 (35.1% OCF-to-revenue) due to accelerating capex, but absolute FCF growing. Balance sheet: $105.5B cash and marketable securities vs $31.7B debt — net cash fortress, no financing risk.11
Form 4 / insider activity: TSM is a Taiwanese-listed company; Form 4 filings are not applicable. Yahoo Finance notes First Eagle Investment reduced its stake in Q1 2026.9 No insider cluster signal available.
Customer concentration: Apple ~20% of revenue, NVDA significant and growing. Hyperscaler direct demand now pulling capacity independently of chip designer intermediaries — structurally reduces single-customer risk.
Pristine. Gross margin at 66%+ and accelerating operating leverage confirm the business quality thesis is tracking the bull case, not the bear.
Entry Price Discipline
B−
Fwd P/E: 25.8× vs sector median ~35× (Semiconductor sector per guide).4 TSM trades at a 26% discount to its own sector — unusual for a monopoly. Residual geopolitical risk premium explains most of this gap.
PEG: 0.53. FY2026 EPS consensus ~$15.66/ADR implies 49% growth over FY2025 ($10.52).2,4 At 25.8× fwd P/E on 49% growth, the multiple is unambiguously cheap on a PEG basis.
Historical context: 3Y avg P/E ~24.5×, 5Y avg ~23.1×, 10Y avg ~21.8×.4 TTM P/E 33.6× is above historical averages, but forward P/E 25.8× is within the historical band given the step-change in earnings growth.
Distance from entry zone: Current ~$410 is 9–21% above my entry zone of $340–375. At entry zone, forward P/E would be 21.7–24.0×, near historical mean — appropriate for a period where structural geopolitical discount has not yet fully resolved.
My variant on EPS path: Consensus FY2026E ~$15.66 may be conservative. Q1 2026 alone was $3.49 and Q2 guidance implies strong sequential growth. If Q2 EPS is $4.0–4.2 and H2 remains strong, FY2026 EPS could reach $16.50–17.00, making forward P/E at $410 closer to 24–25×. This supports the view that valuation is fair-to-reasonable at current prices, but not compelling.
Flow notes: First Eagle (13F, ~45-day lag) reduced TSM position in recent filing.9 Not a meaningful contrarian signal given stock's 120%+ 1Y move — likely profit-taking. IO Fund (Q2 2026) cites Fwd P/E 23.2× against a historical range of 13.5–29.6× — mid-range — and Fwd P/S 11.5× against 5.9–13.4× — slightly above mid-range.11 If post-Q1 EPS consensus has since been revised upward (consistent with guidance raise to "above 30%"), the effective forward multiple may be closer to 23–24× at current prices, making the Fwd P/E of 25.8× shown above (GuruFocus, 13 May 2026) a modestly conservative read.4,11
PEG: 0.53. FY2026 EPS consensus ~$15.66/ADR implies 49% growth over FY2025 ($10.52).2,4 At 25.8× fwd P/E on 49% growth, the multiple is unambiguously cheap on a PEG basis.
Historical context: 3Y avg P/E ~24.5×, 5Y avg ~23.1×, 10Y avg ~21.8×.4 TTM P/E 33.6× is above historical averages, but forward P/E 25.8× is within the historical band given the step-change in earnings growth.
Distance from entry zone: Current ~$410 is 9–21% above my entry zone of $340–375. At entry zone, forward P/E would be 21.7–24.0×, near historical mean — appropriate for a period where structural geopolitical discount has not yet fully resolved.
My variant on EPS path: Consensus FY2026E ~$15.66 may be conservative. Q1 2026 alone was $3.49 and Q2 guidance implies strong sequential growth. If Q2 EPS is $4.0–4.2 and H2 remains strong, FY2026 EPS could reach $16.50–17.00, making forward P/E at $410 closer to 24–25×. This supports the view that valuation is fair-to-reasonable at current prices, but not compelling.
Flow notes: First Eagle (13F, ~45-day lag) reduced TSM position in recent filing.9 Not a meaningful contrarian signal given stock's 120%+ 1Y move — likely profit-taking. IO Fund (Q2 2026) cites Fwd P/E 23.2× against a historical range of 13.5–29.6× — mid-range — and Fwd P/S 11.5× against 5.9–13.4× — slightly above mid-range.11 If post-Q1 EPS consensus has since been revised upward (consistent with guidance raise to "above 30%"), the effective forward multiple may be closer to 23–24× at current prices, making the Fwd P/E of 25.8× shown above (GuruFocus, 13 May 2026) a modestly conservative read.4,11
Above entry — wait for pullback. At $410 (near 52W high), the risk/reward for a new allocation is asymmetric against. The thesis fully justifies a 30–35× fwd P/E on geopolitical de-risking, but I'd want to build position at the historical average multiple, not above it.
Verdict — Add (pullback)
TSM is the single most important company in the AI hardware supply chain — a monopoly at leading-edge foundry with confirmed margin expansion, 8 consecutive earnings beats, and a geopolitical discount that is structurally (if slowly) compressing via the April 2026 US-Taiwan tariff framework. The thesis is clean and Dim 1 + Dim 2 are both A-grade. The constraint is purely entry price: at ~$410, the stock is at its 52W high, 9–21% above my entry zone of $340–375, and above both DCF scenarios on a clean FCF basis. The forward P/E of 25.8× on 49% EPS growth is optically cheap (PEG 0.53) and the semiconductor sector discount is real — but I don't want to initiate or add at 52W high. A pullback to $340–375 puts the stock at 21–24× forward earnings — at or below the 5Y historical mean — and would represent a compelling entry. Do not add above $400.
New position: size to 8–10% of total portfolio at entry zone $340–375. (Starter allocation of 2–3% acceptable above $375 but below $400 if you want exposure before the pullback.)
Opportunity-cost check
▾
vs CSPX
TSM offers a concentrated bet on the leading-edge foundry monopoly at the heart of the AI infrastructure buildout — CSPX gives 0.7% exposure to TSM and blends it with irrelevant sector weight; the variant perception (geopolitical discount compression) is zero-weight in an index. Edge: structural alpha from mispriced geopolitical risk premium that CSPX cannot capture.
vs current holdings
TSM is pure-play foundry with no direct overlap with MAGS (software/hyperscaler weight), PLTR (software), or IBIT (crypto). It stacks naturally with MU (both are AI hardware supply chain, different layers — memory vs logic). A TSM position would complement, not duplicate, existing holdings. If sizing to 8–10%, the displacement question is which current position is least conviction — likely trim MAGS if it is over-indexed to software vs hardware.
Thesis-breakers
▾
- 01TSM N3/N2 utilisation falls below 80% for any quarter through Q4 2027 — signals broad AI capex pullback or customer inventory correction, kills the demand permanence thesis.
- 02Gross margin declines sequentially for two consecutive quarters below 60% — signals that advanced-node pricing power is eroding (Samsung/Intel Foundry closing yield gap) or that Arizona cost inflation is structural, not temporary.
- 03AI revenue CAGR guidance reduced below 30% (from current >mid-40%) at any earnings call through 2027 — direct management admission that the AI infrastructure super-cycle is truncating faster than anticipated.
Key risks
▾
- 01Taiwan geopolitical tail: A kinetic conflict scenario or Chinese blockade affecting TSMC's Hsinchu/Tainan operations would be existential. Market assigns ~5–10% probability to meaningful disruption over a 5Y horizon; even if partly hedged by Arizona, >80% of leading-edge capacity remains Taiwan-based through 2030. Magnitude: 70–90% drawdown in disruption scenario. Not a base case but a fat tail.
- 02AI capex deceleration: If large hyperscalers (MSFT, META, GOOGL, AMZN) cut 2027 capex budgets by 20–30% due to ROI doubts on AI, TSMC's utilisation falls and ASPs compress. 2024 experience (post-ChatGPT demand normalisation) showed how quickly growth expectations can reset. At 30× TTM P/E, a miss of 1–2 quarters of revenue guidance could deliver a 30–40% drawdown.
- 03Arizona cost drag: TSMC's US fabs reportedly run 50% higher cost per wafer than Taiwan equivalents due to construction, labour, and supply chain costs. If Arizona scales faster than revenue, gross margin expansion could stall or reverse by 2027–2028, before operating leverage catches up. 2026 capex at $54B already compresses FCF margin to ~28–30% vs a structurally cleaner ~35%+.
- 04USD/TWD FX headwind: TSMC reports in NTD; ~80–85% of revenue is USD-denominated. USD weakness vs NTD expands the gap between USD-reported and NTD-reported numbers. Q1 2026 revenue was 35.1% YoY in NTD vs 40.6% in USD — if USD weakens further, NTD-reported results look weaker and dividend yield in USD shrinks. Not a thesis-killer but a 5–10% valuation headwind in appreciation scenarios.
- 05Inference efficiency compression: DeepSeek-style breakthroughs compressing compute-per-query requirements could reduce AI accelerator demand per unit of AI output. TSMC's bull case assumes AI scaling laws hold and demand for training clusters continues. A structural efficiency shift that saturates compute demand faster-than-expected is a FY2028+ downside risk to the AI CAGR narrative.
DCF scenarios
▾
Discount rate: 11% — High-growth profitable tech (TSMC risk profile per guide; Taiwan geopolitical exposure warrants the upper end of 10–12% range). FCF base year: FY2025. FY2025 FCF = operating CF NT$2.3T (~$73B) less capex $40.9B = $32.1B. Shares: 5.19B ADR-equivalent. Terminal growth: 5%.
5Y FCF CAGR
32%
Terminal growth
5%
Fair value / ADR
~$297
vs ~$410
−28%
Bull path: FCF grows at 32% p.a. (matching 2025 revenue growth trajectory) as N2/A16 ramp drives ASP uplift and CoWoS packaging volume scales. Y1–Y5 FCFs of $42.4 → $56.0 → $73.9 → $97.6 → $128.8B. Total PV of FCFs ~$293B + terminal value PV ~$1,247B = intrinsic value ~$1,540B ÷ 5.19B shares = $297/ADR. FCF compounding materialises only if (a) AI capex cycle sustains 5 years and (b) Arizona cost drag does not materially offset Taiwan margin expansion.
5Y FCF CAGR
22%
Terminal growth
5%
Fair value / ADR
~$208
vs ~$410
−49%
Bear path: 30% deceleration immediately (22% CAGR) as AI capex normalises post-2026, Arizona cost drag compresses FCF margin to 20–22%, and Samsung closes yield gap at N2 by 2028 forcing moderate price concessions. Y1–Y5 FCFs of $39.2 → $47.8 → $58.3 → $71.1 → $86.7B. Total PV of FCFs ~$226B + terminal value PV ~$856B = intrinsic value ~$1,082B ÷ 5.19B = $208/ADR.
Position vs DCF: Current price ~$410 is above both bull ($297) and bear ($208) scenarios on a clean FY2025 FCF basis. This is not unusual for a monopoly compound — the market is pricing a longer high-growth runway than a 5Y DCF can fully capture. However, it does mean there is no DCF margin of safety at $410. The forward P/E of 25.8× and PEG of 0.53 are the better valuation anchors for this name, and both support the "fair-to-cheap" conclusion. Reconciliation: the $32B FCF base is understated for 2026 — if Q1 2026 run-rate FCF of $8B+ is annualised, the base is ~$35–40B which compresses the DCF discount modestly.
Catalyst timeline
▾
-
2026-07Q2 2026 earnings (~16 Jul): Guidance implies $39.0–40.2B revenue — a ~12% QoQ acceleration. If gross margin holds 65.5–67.5% as guided, this would be the strongest sequential revenue quarter in TSMC's history and could re-rate the stock through the $420 resistance level. A miss here (especially on margin) triggers the trim-entry discussion.
-
2026-Q3NVDA Rubin ramp (H2 2026): NVDA Rubin/Vera Rubin NVL288 enters volume production at TSMC N3. TSMC confirms Rubin in mass production → direct validation of continued HPC demand absorption; likely positive earnings guide revision for Q4 2026 and FY2027.
-
2026-Q4Arizona P2 fab equipment move-in (Oct 2026): TSMC's second Arizona fab begins tool installation, targeting mini-line Q2 2027 and volume Q4 2027. Successful milestone execution partially de-risks the geopolitical discount and confirms US-capacity timeline is on track — structural re-rating catalyst.
-
2027-H1N2 volume ramp visibility: TSMC's N2 (2nm-class) node enters volume production for iPhone 18 / NVDA Rubin Ultra platform. First clear revenue contribution from N2 and ASP data vs N3 will confirm (or deny) the pricing power thesis at the next process node — single most important indicator for 2027 margin trajectory.
Named analyst commentary
▾
"TSMC is one of the least sensational management teams in the AI stocks space, yet management explicitly called AI a multi-year 'megatrend' in their most recent earnings call, with demand now being pulled not just by chip designers, but directly by hyperscale cloud providers seeking to lock in capacity. Perhaps most importantly, TSM is not a 'flip the switch' business model to where demand can be turned on and turned off quickly."
— Beth Kindig / IO Fund (@bethkindig), io-fund.com, 29 Jan 2026
"TSMC's ability to generate exceptionally strong profits showcases that the company is one of the best-managed companies in the world. Despite the rising inflation, tariff concerns, technological advancement, trade wars, overseas fab expansion, and geopolitical tensions, TSMC has overcome these challenges by continuing to generate superior profits. … Nvidia would be locking up more than 50% of [CoWoS] current supply at 650K [wafers], with AMD getting less than 7%."
— IO Fund, "Top 15 Stocks for Q2 2026 — TSMC: The Importance of CoWoS Capacity", io-fund.com, Q2 2026
"[Hyperscalers] went to TSMC and said 'We want more. We want more.' TSMC replied, 'Sorry guys, we're sold out. We can maybe get 5–10% more for 2026, but really we're going to work on 2027.' … Where the real bottleneck is: from CoWoS shortages in 2023 to data centers and power in 2024–25 back to semiconductor fabs in 2026 and beyond."
— Dylan Patel / SemiAnalysis (@SemiAnalysis), latent.space interview, 2026-03
Serenity (@aleabitoreddit): No direct TSM-specific quote found in searched sources. Serenity's bottleneck framework positions TSM as the Chips/Foundry layer constraint — consistent with Dylan Patel's fab-scarcity characterisation above.
Sources
1 SEC EDGAR — TSMC Form 6-K (Q1 2026 earnings release), filed 2026-04-16 · EDGAR link
2 Tickeron / Investing.com — Q1 2026 earnings call transcript, accessed 2026-05-13
3 SEC EDGAR — TSMC Form 6-K (Q4 2025 earnings release + FY2025 results), filed 2026-01-15
4 GuruFocus — Forward P/E 25.81 as of 2026-05-11; FinanceCharts — TTM P/E 30.04, Fwd P/E 24.33; FullRatio — TTM P/E 38.64, 10Y avg 21.83. ⚠ DATA NOTE: TTM P/E varies across sources (30–38×) due to NTD/USD translation timing and ADR share conversion. Using GuruFocus TTM 33.65× / Fwd 25.81× as primary.
5 IO Fund / Beth Kindig — "The Future of AI Stocks? TSMC Commentary Suggests AI Megatrend", io-fund.com, 2026-01-29
6 CNBC / Financial Times — US-Taiwan tariff deal, Taiwanese $250B US investment commitment, 2026-01-15; TSMC Arizona $465B expansion framework, 2026-04-27
7 Dylan Patel / SemiAnalysis — latent.space interview transcript, 2026-03
8 Tom's Hardware / EE News Europe — TSMC Arizona P2 fab timeline and $465B buildout plan, 2026-02 / 2026-04
9 Yahoo Finance — First Eagle Investment stake reduction noted in institutional activity summary, 2026-Q1 (13F data, ~45-day lag)
10 SEC EDGAR — TSMC April 2026 monthly revenue report (NT$410.7B, +17.5% YoY), filed 2026-05-08
11 IO Fund / Beth Kindig — "The IO Fund's Top 15 Stocks for Q2 2026: TSMC — The Importance of CoWoS Capacity", io-fund.com, Q2 2026 (paywalled)
⚠ = figure not cross-verifiable or sourced from single provider. Semiconductor sector fwd P/E median of ~35× per investment-research.md guide (May 2026 reference).
1 SEC EDGAR — TSMC Form 6-K (Q1 2026 earnings release), filed 2026-04-16 · EDGAR link
2 Tickeron / Investing.com — Q1 2026 earnings call transcript, accessed 2026-05-13
3 SEC EDGAR — TSMC Form 6-K (Q4 2025 earnings release + FY2025 results), filed 2026-01-15
4 GuruFocus — Forward P/E 25.81 as of 2026-05-11; FinanceCharts — TTM P/E 30.04, Fwd P/E 24.33; FullRatio — TTM P/E 38.64, 10Y avg 21.83. ⚠ DATA NOTE: TTM P/E varies across sources (30–38×) due to NTD/USD translation timing and ADR share conversion. Using GuruFocus TTM 33.65× / Fwd 25.81× as primary.
5 IO Fund / Beth Kindig — "The Future of AI Stocks? TSMC Commentary Suggests AI Megatrend", io-fund.com, 2026-01-29
6 CNBC / Financial Times — US-Taiwan tariff deal, Taiwanese $250B US investment commitment, 2026-01-15; TSMC Arizona $465B expansion framework, 2026-04-27
7 Dylan Patel / SemiAnalysis — latent.space interview transcript, 2026-03
8 Tom's Hardware / EE News Europe — TSMC Arizona P2 fab timeline and $465B buildout plan, 2026-02 / 2026-04
9 Yahoo Finance — First Eagle Investment stake reduction noted in institutional activity summary, 2026-Q1 (13F data, ~45-day lag)
10 SEC EDGAR — TSMC April 2026 monthly revenue report (NT$410.7B, +17.5% YoY), filed 2026-05-08
11 IO Fund / Beth Kindig — "The IO Fund's Top 15 Stocks for Q2 2026: TSMC — The Importance of CoWoS Capacity", io-fund.com, Q2 2026 (paywalled)
⚠ = figure not cross-verifiable or sourced from single provider. Semiconductor sector fwd P/E median of ~35× per investment-research.md guide (May 2026 reference).