Taiwan Semiconductor (TSMC) Posted Record Profit and the Stock Still Fell. Here's What the Market Is Worried About.

TSMC reported Q3 revenue up 50% and net profit up 69% year over year, both records, with HPC revenue rising 22% sequentially and accounting for 68% of sales. The company raised its full-year USD revenue growth outlook to 44% from at least 40%, while holding its $60B-$64B capex target; gross margin eased to 66.4% from 67.7% in Q2. The article cites heavy investment, potential AI-spending moderation, and prior share-price gains as reasons for the stock pullback, but views TSMC as attractive for long-term investors at 16 times next year's earnings.

Analysis

The key debate is cash conversion, not demand: accelerating advanced-node mix can support pricing and utilization, but the fixed capex plan plus overseas-fab and 2nm ramp costs may delay free-cash-flow leverage. A sequential gross-margin decline is tolerable if temporary; it becomes a valuation problem if depreciation and ramp inefficiency keep margins below the prior quarter even as revenue grows. Watch gross margin, 2nm yields/utilization, and capex-to-revenue—not headline EPS alone.

Second order: TSMC capacity remains a common bottleneck for Nvidia, AMD, and Broadcom, so any production shortfall can constrain their product shipments, while successful capacity expansion helps all three compete for AI workloads. Conversely, Amazon or Microsoft capex restraint would hit TSMC through several customers’ demand plans, not just one chip vendor. Taiwan-related disruption is a low-frequency, high-severity tail risk that diversification across customers does not hedge.

Contrarian view: a pullback after strong results need not signal weakening AI demand; investors may be discounting near-term margin/FCF dilution before the new-node capacity earns returns. But the article’s growth forecasts and valuation framing are not independently established here, so do not underwrite the thesis on those alone. Over 1–3 months, hyperscaler spending signals and quarterly margin guidance dominate; over 6–18 months, 2nm ramp economics and capex returns decide whether growth translates into per-share cash generation.

Key Decisions for Investors

• Consider scaling into TSM on weakness rather than chasing the earnings reaction; size it as an AI-capex exposure, not a low-risk diversified semiconductor holding. Favor staged entries until margin direction confirms that the ramp is manageable.
• For a 1–3 month catalyst check, track Amazon and Microsoft capex commentary plus TSM’s next gross-margin guidance and 2nm utilization/yields. A demand slowdown or repeated sequential margin deterioration would falsify the constructive case; stronger spending and stabilizing margins would support adding.
• Avoid an unhedged short in Nvidia, AMD, or Broadcom solely because TSM’s margin dipped: their near-term exposure is product demand and available foundry capacity, and the article does not establish a customer-specific slowdown. Broad semiconductor exposure can be reduced with a sector hedge if AI-capex risk rises.
• Reassess over 6–18 months against free-cash-flow conversion and returns on the elevated buildout. If revenue growth continues but margins and cash generation fail to recover as new nodes ramp, reduce TSM; Taiwan disruption remains a separate tail-risk trigger rather than a routine earnings catalyst.