⚠ The Capital Burden Falls on TSMC TooModerate threat

Capital intensity is a moat, but the moat is paid for out of TSMC's own pocket, and the bill keeps rising. Capital spending climbing toward sixty-four billion dollars a year¹ is an enormous, escalating claim on the company's cash flow: by TSMC's own count, capital spending in the five years to 2025 totalled $167 billion and R&D another $30 billion², and each new node costs more to develop and build than the last. In a strong-demand environment the spending is easily funded and richly rewarded, but the sums are so large that they materially constrain free cash flow, and a misjudgment of demand would leave TSMC having sunk tens of billions into capacity it cannot fill.

The overseas expansion sharpens the burden. Building fabs in Arizona, Japan, and Germany — driven by geopolitics and customer/government pressure rather than pure economics — costs substantially more than building in Taiwan, at lower initial yields and higher operating costs, and TSMC has acknowledged these overseas fabs will dilute its gross margins for years. So the capital burden is not only rising but being partly directed toward less economic locations for strategic reasons. TSMC's scale and cash generation make the spending affordable, and it has always translated capital into returns better than any peer. But the capital intensity that keeps competitors out is a heavy and growing weight on TSMC itself — a moat that must be continuously rebuilt at ever-greater expense, and one whose cost is now inflated by the geopolitics driving expensive diversification.