Dossier · TSM · Held
TSM · Taiwan Semiconductor Manufacturing Company Ltd.
Last analysed ·
Current thesis
TSM — held research dossier: thesis, invalidation trigger, archetype.
Invalidation trigger
a daily close below the published invalidation level
Thesis status
Open commitment scored if the trigger above fires How this is scored →Current Thesis
The foundry-monopoly story is shifting from demand confirmation to pricing capture. The April Q1 print (rev $35.90B vs $35.50B est, EPS $3.49 vs $3.31, Q2 guide midpoint ~$39.6B = +38% YoY, reported 2026-04-16) settled the demand leg. The 06-01/06-02 reports of a 15% price hike on 3nm for 2026 with further hikes flagged for 2027 open the pricing leg. The cost base does not rise 15% in lockstep, so that increment drops toward gross margin as scarcity rent, stacked on a 3nm mix already carrying ~25% ASP premium to 5nm. CEO C.C. Wei reinforced it on 2026-06-04, calling AI capacity constrained "for a very long time" and hinting at additional hikes. Theme is ACCELERATING sell-side narrative is now rotating explicitly toward AI suppliers (2026-05-29). The friction: price sits near 52-week highs and took a -2% profit-taking dip on 2026-06-04, so the easy money on the April gap is paid. Next hard read is the May monthly revenue release ~2026-06-10.
Bull Case
- 2026-06-01/02 15% 3nm price hike for 2026, more flagged for 2027. Cost base doesn't move 15%; the increment is near-pure gross-margin rent layered on a 3nm mix already at ~25% ASP premium to 5nm.
- 2026-06-04 CEO Wei: AI capacity tight "for a very long time," hints at further hikes. Frames the supply-demand gap as multi-year and structural, not a single-cycle squeeze.
- 2026-06-04 CEO: rivals have "years to catch up." Operator reaffirms the leading-edge lead exactly when Intel 18A / Samsung 2nm are the live bear risk.
- 2026-06-03 Intel CEO Lip-Bu Tan calls TSM "trusted partner, not rival," cites surging agentic-AI CPU demand. The single most-cited defection name reframes as a customer, defusing the near-term 18A-poaching script.
- 2026-06-01 Nvidia RTX Spark AI-PC chip names TSM a beneficiary. AI-PC is a fresh volume vector layered onto Blackwell/Rubin, AMD MI350/MI355X and Apple M-series.
- 2026-06-01 TSMC uses NVIDIA AI to speed its own chip design/manufacturing. Customer becomes tooling partner; the NVDA-TSM symbiosis deepens on both sides of the invoice.
- 2026-05-29 "AI's next winners may be its suppliers, not Nvidia's buyers." Sell-side rotation toward the supply chain is early, and TSM is the prime instance.
- 2026-06-05 semiconductor dip bought aggressively (Momo); blowout jobs report. Dip demand for semis is intact and a risk-on macro tape supports the high-beta complex.
Bear Case
- 2026-05-29 / 06-02 Dan Loeb (Third Point) cut TSM and ~90% of NVDA in Q1, added Broadcom. A marquee semi bull rotating out of foundry and into custom silicon (AVGO) signals AI-premium leadership may be migrating to ASIC names.
- 2026-06-01 the same 15% hike pressures Nvidia/AMD/Apple margins. A TSM positive near-term, but sustained hikes raise the incentive for top customers to dual-source (Samsung 2nm, Intel 18A) or accelerate in-housing the one risk the thesis cannot absorb.
- 2026-06-02/06-04 price at 52-week highs, -2% profit-taking dip. Names well above the 200-day SMA tend to give back 5-10% on the first wobble when no fresh catalyst lands.
- Taiwan-Strait tail is structural and unhedgeable. Any PLA escalation gaps the equity 15-25% instantly; this caps prudent conviction regardless of narrative quality.
- Overseas-fab dilution (Arizona, Kumamoto). 20-30% higher cost; the overseas mix grows through 2026-27 and drags blended gross margin against the pricing tailwind.
Setup & Price Structure
Confirmed uptrend, trading well above the 200-day SMA (2026-06-01), with the post-04-16 earnings gap as the launch base. Price made fresh 52-week highs into 06-02, then took a -2% profit-taking dip 06-04 orderly, not distribution. Sell-side has been lifting targets: Needham $480 (2026-04-16), Barclays Overweight PT $470 (2026-04-22) an active upgrade cycle. The caution is technical: RSI ran hot into the highs, and entries chasing a name this stretched above its rising 20-EMA are the ones that give back first. A clean entry zone is a pullback to the rising 20-EMA / breakout shelf rather than a chase into the high. A daily close that loses that rising 20-EMA / April breakout base is the structural break.
Catalyst Calendar (next 30 days)
- ~2026-06-10 (est.) May monthly revenue release. Highest-signal interim catalyst; the YoY/MoM trajectory is the direct read on whether the pricing-power second leg is reaching the topline.
- 2026-06 (ongoing) Computex / Jensen Huang keynote follow-through (2026-06-01). AI capex and Taiwan-investment commentary feed the demand narrative.
- ~2026-07-10 (est.) June monthly revenue release (just outside window; completes the Q2-quarter read).
- ~2026-07-17 (est.) Q2 earnings print. Outside the 30-day window. Binary risk; fresh positioning inside T-3d (~from 2026-07-14) carries earnings-gap exposure.
What Would Change Our Mind
The whole bull case is the leading-edge monopoly plus pricing power. It breaks if: (1) any top-5 customer (Apple, Nvidia, AMD) confirms shifting leading-edge volume to Intel 18A or Samsung 2nm a hard-stop, not a wait-and-see; (2) a monthly revenue print shows sequential decline, signaling the demand/pricing leg is rolling; (3) price closes below the rising 20-EMA / April breakout base, marking the trend break; (4) the custom-silicon rotation accelerates more marquee funds following Loeb out of foundry into ASIC and TSM underperforms AVGO/peers on up days, draining the AI premium multiple. A Taiwan-Strait escalation is the unhedgeable override that exits the thesis regardless of any other signal.
Correlation Notes
TSM sits at the core of the AI-silicon complex high correlation to NVDA, AMD, AVGO and the SOX/SMH semis tape; it leads on capex-demand headlines and lags on custom-silicon-rotation days. The Loeb AVGO-over-TSM rotation (2026-05-29) is the relative-strength pair to monitor: TSM underperforming AVGO on up days flags AI-premium migration toward ASICs. Idiosyncratic Taiwan geopolitical risk means TSM should not be the only Taiwan-foundry expression of the AI-silicon theme; equipment / US-fab names (ASML, AMAT, and US-domiciled fab capacity) carry the theme without doubling the Strait tail. Macro: high-beta and rate-sensitive the 2026-06-05 blowout jobs report cuts both ways (risk-on demand against higher-for-longer discount-rate pressure on long-duration multiples).
Notes
- 2026-04-19: Leading-edge foundry monopolist; all AI chips flow through here
- Earnings blackout: next print ~2026-07-17 (Q2). Do not initiate inside T-3d.
- Taiwan Strait tail-risk is structural and unhedgeable never size past HIGH on TSM alone; pair with US-fab exposure if thesis runs.
- Monthly revenue release is ~10th of each month single highest-signal interim catalyst.
- Foundry-monopoly premise is the whole bull case. Any Intel 18A or Samsung 2nm top-5 customer defection is a hard-stop
- not a 'wait and see'.
- Foundry-monopoly premise is the whole bull case. Any Intel 18A or Samsung 2nm top-5 customer defection is a hard-stop
- not a 'wait and see'.
- 2026-04-22: Barclays Overweight, PT raised to $470 second sell-side PT hike in a week (Needham $480 on 2026-04-16); upgrade cycle live.
- Post-2026-04-16 gap: do NOT chase at RSI>70; wait for 20-EMA retest. Never average down if post-entry tape breaks; re-enter fresh setup only.
- Earnings blackout: Q2 print ~2026-07-17. Do NOT initiate fresh size inside T-3d (~from 2026-07-14).
- Taiwan-Strait tail is structural and unhedgeable never size past HIGH on TSM alone; pair the AI-silicon theme with US-fab/equipment exposure, not a second Taiwan foundry.
- Monthly revenue release ~10th of each month is the single highest-signal interim catalyst; next ~2026-06-10.
- Foundry-monopoly premise is the whole bull case: any top-5 customer defection to Intel 18A / Samsung 2nm is a hard-stop, not wait-and-see.
- 2026-06-01/02: reported 15% 3nm price hike for 2026 (more flagged for 2027) pricing power is the new accelerant; near-pure GM tailwind on top of the April beat.
- the published invalidation level. Don't chase near 52-week highs at RSI>70 add only on a 20-EMA retest.
- 2026-05-29/06-02: Dan Loeb cut TSM + NVDA, added AVGO watch custom-silicon rotation as the AI-premium-multiple tell.
- Earnings blackout: Q2 print ~2026-07-17. Binary risk; no fresh size inside T-3d (~from 2026-07-14).
- Taiwan-Strait tail is structural and unhedgeable cap conviction at HIGH on TSM alone; pair the AI-silicon theme with US-fab/equipment exposure (ASML/AMAT/LRCX), not a second Taiwan foundry.
- 2026-06-01/02: 15% 3nm price hike for 2026 (more flagged 2027) pricing power is the new accelerant; near-pure GM tailwind on top of the April beat.
- 2026-05-29/06-02: Dan Loeb cut TSM + NVDA, added AVGO watch custom-silicon rotation (TSM vs AVGO relative strength) as the AI-premium-multiple read.
- Don't chase near 52-week highs at RSI>70 clean entry is a rising-20-EMA / breakout-shelf retest.
Related · shared themes
ASML
ASML Holding N.V.
ASML is the cleanest second-derivative on the AI-capex super-cycle: EUV monopoly, CEO's "supply-limited market for quite a while" framing (5/20), High-NA first product data on logic AND DRAM later in 2026, JPM Overweight PT $2,200 (6/3). Theme is classifier-ACCELERATING with cluster confirmation and 7 rules fired (momentum top-25%, +19pp vs SPX 20d, near-52w-high structure, RSI 67.6, StockTwits +87%). ASML is the quality expression, not diversification vs TSM. Entry is a vertical near-52w-high, not the pullback-with-volume gate. Avoid: add AI-infra via a non-correlated name, or take ASML only on a 20/50-DMA pullback once the TSM cluster slot frees.
AVGO
Broadcom Inc.
Q2 FY26 (2026-06-03) printed a monster $30B AI bookings vs $10.8B shipped (2.8x book-to-bill), $100B+ FY27 AI guide but AVGO sold the news (-12% on 2026-06-04) and kept falling into a rate-driven risk-off. Fundamentals ACCELERATING, tape CORRECTING; the buyable structure is the higher-low that holds, not the knife.
AMD
Advanced Micro Devices, Inc.
AI-chip ACCELERATING but (rank 34%), parabolic ATH chase, ARK persistent distribution dossier wants the ~$420 shelf retest.
INTC
Intel Corporation
AI-chip theme but momentum LAGGARD (rank 88%), NVDA ARM-PC structural x86 threat + 108x P/E contested thesis, weakest cluster name.