The major bottleneck for AI has always been TSMC.

```

Tech giants are pouring money into acquiring AI chips, putting unprecedented pressure on global chip manufacturing capacity, and the biggest beneficiary of this supply crunch points to the same name—TSMC.

Microsoft, Meta, Alphabet, and Amazon have planned combined capital expenditures of up to $725 billion this year, with most of it flowing into the procurement of AI chips. This surge in demand has directly boosted TSMC’s capacity utilization and gross margin—in the first quarter this year, its gross margin climbed from about 59% a year ago to around 66%. TSMC CEO C.C. Wei said last month that he is "full of confidence" about this year's revenue growth exceeding 30%.

In the perspective of chip investors, memory chip makers, Intel, and AMD have recently been more in the spotlight, but according to analysis by The Wall Street Journal, in this new phase of AI, no chip company has a more structural advantage than TSMC.

What’s even more notable is that despite its strong fundamentals, TSMC’s current stock price trades at a forward P/E ratio of only about 21, below the Philadelphia Semiconductor Index's average of 26, so its valuation is not expensive.

Supply bottlenecks tightening, TSMC collects pricing power

TSMC's core competitiveness lies in its irreplaceable market position. In the field of the most advanced process chips, TSMC has virtually no real competitors—Samsung's foundry business revenue lags far behind, Intel and Japan’s Rapidus are still struggling to gain footing, and even Elon Musk’s recently announced Terafab project, with the help of Intel, is still far from mass production.

This near-monopoly position gives TSMC implicit pricing power amid extremely tight supply and demand. Nvidia’s purchasing commitments for the latest fiscal year ending in January exceeded $95 billion, a substantial proportion of which will be paid to TSMC—just two years ago, this figure was only $16 billion. Some clients even lock in capacity years in advance and pay tens of billions of dollars upfront to secure chips.

Nevertheless, C.C. Wei’s wording to the public is cautious. Last month, he said: “We will not significantly change our pricing, we just ensure our clients can succeed in their respective markets.” This statement both reassures client relationships and leaves room for natural price increases as processes upgrade in the future.

Short-term gross margin pressure, long-term logic unchanged

TSMC CFO Wendell Huang admitted to analysts last month that gross margin will narrow in the second half of this year, because the company is entering the ramp-up stage of mass production for its latest generation N2 process chips. Costs are higher in the early stage of a new process; once production stabilizes, costs will fall accordingly. This is a common rule in the chip industry.

In addition, TSMC's expansion of wafer plants in the United States is also putting pressure on profit margins, since factory operating costs in the US are higher than in Taiwan.

From a longer cycle perspective, as advanced processes mature, gross margins are expected to rebound. TSMC is also expanding N3 process capacity in Taiwan, Japan, and the United States—this generation of technology is slightly less advanced than the cutting-edge process, but required equipment costs are lower and it can still take on large AI chip orders, balancing profitability with capacity expansion.

Capital expenditures remain high, but growth slower than revenue

TSMC’s capital expenditures this year are expected to approach the upper end of the previously forecast range of $52–56 billion. In the chip industry, excessively rapid capacity expansion has always been a risk—once demand drops, idle capacity becomes a heavy fixed cost burden.

But TSMC’s current expansion pace is not out of control. C.C. Wei’s statements provide a key reference: This year's revenue growth is expected to exceed 30%, higher than capital expenditure growth, which means the company is not borrowing from the future.

Current demand visibility is unusually clear. Clients’ long-term purchasing commitments and prepayment arrangements provide TSMC with order lock-in periods far beyond industry norms, significantly reducing uncertainty in capacity investment.

Overall, TSMC boasts high capacity utilization, an almost monopolistic position in advanced processes, and steadily rising client prepayment commitments, yet trades at a valuation below the industry average. About 21 times forward earnings, not only lower than the Philadelphia Semiconductor Index average, but also far below hotter peers like Intel and AMD.

Risk warning and disclaimerThe market carries risks and investments must be cautious. This article does not constitute personal investment advice, nor does it consider the special investment objectives, financial conditions, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular situation. Should you invest accordingly, the responsibility is yours. ```