Is Taiwan Semiconductor Manufacturing Stock a Buy on Strong AI Demand?

TSMC delivered strong Q1 results, with revenue up 35% year over year and gross margin expanding to 66.2% from 58.8%, a 740bps increase. AI chip demand remained the key driver, with high-performance computing revenue rising 55% sequentially and management saying demand still far outpaces supply. The company guided Q2 revenue to $39.0B-$40.2B and maintained full-year revenue growth outlook above 30%.

Analysis

The key second-order read-through is that TSMC is not just capturing AI demand; it is exercising pricing power while still controlling yield enough to keep gross margin expanding. That combination is unusual in semis and implies the bottleneck is migrating from chip design to advanced packaging, leading-edge lithography, and wafers at 3nm/2nm, which should keep foundry pricing firm for at least the next 4-8 quarters.

For NVDA, this is supportive but not equally elastic: stronger TSMC capacity reduces the odds of a supply shock, yet it also confirms that AI demand remains broad enough to absorb incremental supply without clearing pressure. The bigger beneficiary may be the equipment and materials stack over the next 6-18 months, because sustained capex at the high end usually precedes a multi-quarter wave of tool orders, while high utilization and node migration keep service intensity elevated.

The market is likely underappreciating that margin expansion is the real catalyst, not just top-line growth. If management is willing to fund capacity at the peak of demand while preserving margins, the cycle may be longer and less mean-reverting than consensus expects; that argues against fading the stock on valuation alone. The main risk is a 2H25 digestion phase if AI customers accelerate inventory builds too far ahead of true end-demand, which would hit wafer starts before it shows up in headline revenue.

Contrarian angle: the bullish consensus may be overfocusing on NVDA while underweighting the fact that TSMC’s economics improve as node transitions tighten supply discipline. If 2nm ramps cleanly, the company could prove that advanced-node migration is still margin-accretive, which would justify a higher structural multiple for the foundry model than the market currently assigns.

Key Decisions for Investors

• Stay long TSM, but add on pullbacks only: use a 3-6 month horizon with a target of a re-rating toward 27-28x forward earnings if margins hold above 65%; risk is a Q3/Q4 digestion phase that could compress the multiple back to low-20s.
• Buy NVDA only as a relative-value expression versus TSMC strength, not as an outright momentum chase: pair long TSM / short a smaller, more supply-constrained AI beneficiary if available; the thesis is that foundry economics are still the cleaner way to monetize AI scarcity over the next 2-4 quarters.
• Initiate a basket long in semiconductor capex beneficiaries over 6-12 months, especially ASML and AMAT, because sustained high-end fab expansion tends to flow through into tools with a lag; risk/reward improves if TSMC keeps capex at the top of guidance.
• For event-driven investors, consider selling downside puts in TSM into any post-earnings volatility over the next 1-2 quarters; premium should remain rich if the market continues to question margin durability, and the fundamental floor is improving.
• Avoid chasing INTC on this print: TSMC’s margin expansion and node lead reinforce the competitive gap, making any catch-up trade more of a multi-year execution story than a near-term catalyst.