TSMC Q3 2026: Sixth Straight Record Quarter, 2nm at 8% of Wafers — and the Stock Falls Anyway
TSMC reported Q3 2026 (quarter ended September 30) on October 15, 2026: consolidated revenue of NT$1,494.24B (~US$46.72B), up 50.9% YoY and 17.6% QoQ, net income NT$763.06B, up 68.7% YoY — a sixth consecutive record profit quarter — with diluted EPS of NT$29.43 (US$4.60 per ADR vs $4.45 expected) (Vested Finance, TSM market view). Both revenue (+2.3%) and profit (+2.2%) beat LSEG consensus (NT$1.46T / NT$746.3B).
Against its own July guidance: revenue of US$46.72B came in ~2% above the top of the guided US$44.6–45.8B range; operating margin of 58.7% beat the guided 56–58%; gross margin of 66.4% landed above the midpoint of the 65–67% guide. GM was down 130bps QoQ from 67.7% — exactly the 2nm ramp dilution management had flagged: "Compared to second quarter, our third quarter gross margin decreased by 130 basis points sequentially to 66.4%, primarily due to the margin dilution from the initial ramp-up of our 2 nm technology" (Q3 earnings call transcript, Yahoo Finance).
The mix underneath is the concentration story: 2nm reached 8% of wafer revenue (from 3% in Q2), 3nm 33%, 5nm 28% — advanced nodes (≤7nm) now 79% of wafer revenue — and high-performance computing rose to 68% of revenue from 57% a year ago (Vested Finance).
Q4 guidance implies no deceleration: revenue of US$52.3–54.3B (midpoint ~+14% QoQ), gross margin 65–67%, operating margin 56.5–58.5% — management expects margins to hold even as 2nm ramps steeply, and it raised its full-year growth forecast. TSMC also maintained 2026 capex of $60–64B (Reuters). News reports put the fab pipeline near 20 new fabs, mostly in Taiwan, and TSMC is reportedly eyeing a Texas site connected with Elon Musk's Terafab project — which could also emerge as a significant ASML customer — after the Arizona expansion (TSM market view news).
The market reaction is the story's second half: despite the beat-and-raise, TSM closed at $412.61, down 9.0% over five days, 15.1% below its 52-week high, RSI 32 (TSM market view). Forbes: "TSMC and ASML delivered the kind of numbers that should have lifted their shares. The response was weak. When a stock stops rewarding an earnings beat, investors are usually worried about something that has not reached the reported numbers yet1" — the worry being the financing cost of customers' capex, not TSMC's P&L (Forbes, The Credit Repricing: CoreWeave's Bonds Go Distressed and the AI Selloff Escapes the Neoclouds).
What it means: the supply side of the AI capex cycle is executing flawlessly and still guiding up — the constraint story is intact (2nm ramping faster than expected, HPC share climbing, capex held at $60–64B). But equity investors are now discounting the financing layer and customer concentration (HPC at 68% of revenue) rather than the foundry's own numbers. Q4's US$53.3B midpoint (+14% QoQ) is the next proof point; any hyperscaler capex wobble at the Oct 21–29 earnings prints lands directly on this trajectory (Hyperscaler Capex: Five-Way 2026 Total Now ~$800B; Cloud Revenue Growth Is the Counterweight to Funding-Layer Stress, The Credit Repricing: CoreWeave's Bonds Go Distressed and the AI Selloff Escapes the Neoclouds).
-
An instance of The AI buildout now trades on next year's capex, not this year's beats. — A sixth consecutive record quarter plus a raised guide still leaves TSMC 15% below its high — the trade now prices customers' future capex financing, not the foundry's P&L. ↩︎