TSM

Current Thesis

Foundry-monopoly pricing power is crystallizing from narrative into invoices. The April Q1 beat (rev $35.90B vs $35.50B, EPS $3.49 vs $3.31; Q2 guide mid $39.6B = +38% YoY) established demand; the June flow establishes PRICING. Reports of a 15% 3nm price hike for 2026 (2026-06-01/02) plus CEO C.C. Wei warning AI capacity stays constrained "for a very long time" and hinting at further hikes (2026-06-04) means scarcity rent is being captured at the wafer line — a near-pure gross-margin tailwind, since TSM's cost base does not rise 15% to match. This is the SECOND leg of the thesis, not a re-rate of the first. Theme ACCELERATING. The catch: tape is extended near 52-week highs (2026-06-02), well above the 200-day SMA, and took a -2% profit-taking dip on 2026-06-04. We hold the runner from [entry redacted]; fresh size only on a 20-EMA pullback, never a chase into the high. Next tape event: May monthly revenue ~2026-06-10.

Bull Case

• 2026-06-01/02 — 15% 3nm price hike reported for 2026 (further hikes flagged for 2027). TSM's cost base does not move 15%; this is monopoly rent dropping ~straight to gross margin, stacked on the 3nm mix shift (25% of wafer rev, ~25% ASP premium to 5nm). Pricing power, the whole bull case, made real.
• 2026-06-04 — CEO C.C. Wei: AI capacity constrained "for a very long time," hints at price hikes ("we still need to..."). Supply-demand imbalance framed as multi-year, not cyclical — scarcity compounds.
• 2026-06-04 — CEO: rivals have "years to catch up." Moat reaffirmed by the operator; the leading-edge monopoly premise intact.
• 2026-06-03 — Intel CEO Lip-Bu Tan calls TSM "trusted partner, not rival," cites surging CPU demand from agentic AI. The marquee defection-risk name reframes itself as a CUSTOMER — kills the Intel-18A-poaching script near-term.
• 2026-06-01 — Nvidia RTX Spark AI-PC chip names TSM a beneficiary; AI-PC is a fresh demand vector stacked onto Blackwell/Rubin, AMD MI350/MI355X, Apple M-series.
• 2026-06-01 — TSMC uses NVIDIA AI tech to accelerate its own chip design/manufacturing — deepening the NVDA-TSM symbiosis; the customer is now also a tooling partner.
• 2026-05-29 — "AI's next winners may be its suppliers, not Nvidia's buyers." Media/sell-side narrative rotating explicitly TOWARD the supply chain — TSM is the prime instance, and the rotation is early, not exhausted.

Bear Case

• 2026-05-29 / 06-02 — Dan Loeb (Third Point) reduced TSM and dumped ~90% of NVDA in Q1, added Broadcom. A marquee semi bull rotating OUT of TSM and INTO custom silicon (AVGO) is a real smart-money tell; AI-premium leadership may be migrating to ASIC names.
• 2026-06-01 — the 15% price hike pressures Nvidia/AMD/Apple margins. Near-term pure TSM positive, but sustained hikes incentivize customers to dual-source (Samsung 2nm, Intel 18A) or accelerate in-housing — the long-run defection risk the thesis cannot survive.
• 2026-06-02/04 — tape at 52-week highs, -2% profit-taking dip. Easy money on the April gap is paid; late entries on a name well above its 200-day SMA tend to give back 5-10% on the first wobble absent a fresh catalyst.
• Taiwan geopolitical tail is structural and unhedgeable — any PLA escalation = instant 15-25% gap. Sizing capped at HIGH on TSM alone regardless of narrative quality.
• Arizona/Kumamoto overseas-fab dilution runs 20-30% higher cost; overseas mix grows through 2026-27 and partially offsets the 3nm/price-hike margin tailwind.

Setup & Price Structure

• Trading near 52-week highs (2026-06-02), well above the 200-day SMA (confirmed 2026-06-01 "strong bullish trend"). Structure intact, uptrend unbroken.
• [position redacted] from 2026-04-28 @ [entry redacted]; tranche-add limits ran [entry redacted] (5/22) → [entry redacted] (5/26), so [entry redacted] currently near the upper end.
• 2026-06-04 -2% premarket = profit-taking, not distribution — needs to hold the 20-EMA (~[entry redacted]) on a weekly-close basis to stay long.
• RSI reset healthy to 54 on 2026-05-21; likely back into the 60s-70s near highs now (no live print supplied). For arch-2 this is confirmation, not an exit signal — RSI>88 + structural cracks would be the trim flag, not reached.
• Hard stop [stop redacted] (close-based), ~7% below entry, ~11% below recent highs.

Catalyst Calendar (next 30 days)

• ~2026-06-10 (est.) — May monthly revenue release. Highest-signal interim catalyst; first order-book read after the price-hike headlines. Watch YoY: Q2 guide implies ~+38%; a print decelerating below +30% YoY is a yellow flag.
• Ongoing through June — 3nm 15% price-hike implementation. Any confirmation/quantification (or visible customer pushback) moves the GM narrative.
• Computex aftermath — Jensen Huang keynote (covered 2026-06-01) and AI-PC chip ramp commentary keep the demand-vector story live.
• NOT in window: Q2 earnings ~2026-07-17. Blackout starts ~3 trading days prior (~07-14). Do not initiate fresh size inside T-3d. Clear until mid-July.

What Would Change Our Mind

• a daily close below the thesis-invalidation level (hard stop) or a weekly close below the 20-EMA (~$390) → structure broken, exit the runner, no averaging down.
• Any top-5 customer (NVDA/AMD/AAPL/AVGO/Qualcomm) defection to Intel 18A or Samsung 2nm → hard-stop, not "wait and see." The monopoly premise cracking.
• May revenue (~06-10) YoY decelerating below +30% → demand-side crack ahead of the Q2 print; cut to probe or exit.
• Theme flip to SATURATED (CNBC-cover mainstream + retail euphoria + sell-side PTs all clustered behind price) with no fresh pricing/capacity catalyst → trim into strength.
• PLA Taiwan-Strait escalation → unhedgeable gap; this is why size never exceeds HIGH.

Correlation Notes

• TSM is the diversified-customer proxy for the entire AI-silicon complex: NVDA (Blackwell/Rubin), AMD (MI350/355X), AVGO/Google (TPU), AAPL (M-series), Tesla, plus AI-PC (2026-06-01). Customer diversification AWAY from any single name (e.g., Google/Marvell TPU) concentrates MORE foundry demand here, not less.
• High beta to SMH/SOXX — moves with the broad semi tape; do not double up TSM with another leading-edge-foundry exposure (there is no true peer, which is the point).
• Watch the Loeb rotation: if smart money keeps migrating NVDA→AVGO (custom silicon), TSM still wins both sides since both fab here — but AVGO relative strength is the tell for where the AI-premium multiple is pooling.
• Pair-risk with the Taiwan tail: any TSM long is implicitly short Taiwan-Strait stability. If sizing up the AI-silicon theme, prefer pairing with a US-fab or equipment name, not a second Taiwan-concentrated foundry.