Giant Taiwanese chipmaker TSMC confirms it’s in talks with Elon Musk to join ‘Terafab’ complex in Texas: ‘The AI market continues to be very strong’

TSMC reported record July–September net profit of NT$763.1 billion (nearly $24 billion), up 69% year over year, and revenue of NT$1.49 trillion ($46 billion), up 51%, as AI demand remained stronger than its capacity to supply. It is evaluating Texas expansion and discussions with Elon Musk’s Terafab project, while saying any Texas plant run by TSMC would need to be operated by TSMC; Terafab’s structure remains unresolved. TSMC raised its 2025 capital-spending budget to $60 billion–$64 billion from $52 billion–$56 billion and expects 2026–2028 spending to be significantly higher, while flagging macroeconomic pressure on consumer- and price-sensitive segments.

Analysis

The key signal is not the profit print but TSMC’s willingness to accelerate capacity while explicitly warning that consumer- and price-sensitive demand is weaker. That raises the chance of a two-speed wafer market: AI accelerators retain priority and pricing power, while weaker end markets absorb higher component costs. NVDA is a direct potential beneficiary if added capacity converts into usable leading-edge output; Apple may face less favorable economics if component inflation reaches devices without matching consumer demand. Neither outcome is guaranteed by a capacity announcement.

For TSMC, higher capex can extend the AI growth runway but also increases execution, utilization and depreciation risk. Texas is not yet a committed project: the difference between TSMC operating a fab and Musk’s description of TSMC subleasing space is material to process control, economics and whether the site can produce advanced chips at scale. Verify ownership, operating responsibility, subsidies, customer commitments and node/yield plans before assigning value to it.

Near term, strong results support sentiment, but the stock may already reflect AI optimism; avoid chasing a post-results gap. Over 1–3 months, watch capex guidance, capacity allocation and customer delivery commentary. Over 6–18 months, returns depend on ramp speed and utilization, not announced spending. The contrarian risk is treating capacity expansion as unambiguously bullish: if AI demand normalizes while consumer weakness persists, added fixed costs could pressure returns. Thesis weakens if TSMC trims its forward investment plan or reports deteriorating profitability/utilization as capacity ramps.

Key Decisions for Investors

• TSM: Prefer staged exposure on post-earnings weakness rather than chasing. Reassess if forward capex is reduced, AI customer demand/delivery commentary weakens, or profitability deteriorates as new capacity comes online.
• NVDA: Treat TSMC capacity expansion as a conditional medium-term supply-chain positive, not immediate incremental revenue. Track evidence that leading-edge output and packaging availability are improving; a persistent delivery bottleneck would invalidate the read-through.
• AAPL: Monitor component-cost commentary and demand indicators. Avoid assuming TSMC’s AI-driven strength transfers to Apple; consumer price sensitivity could limit pass-through and leave device margins or volumes exposed.
• Texas/Terafab: Keep the project out of base-case valuation until TSMC and Musk reconcile operating control and disclose economics, timing, process capability and committed customers. A mismatch on who runs the fab is a meaningful execution and governance risk.