⚠ The Lock-In Is Mutual, and the Buyers Are MightyModerate threat
TSMC (TSM) — threat to the moat
A handful of giants depend on TSMC — and giants negotiate like giants.
The customer moat's central caveat is that the lock-in runs both ways, and the buyers on the other end1 are among the most powerful companies in the world. TSMC's leading-edge revenue is concentrated in a small number of giants — Apple, Nvidia, AMD, the hyperscalers — and while they depend on TSMC, TSMC depends on them just as heavily, because they are the only buyers able to afford and consume its most advanced capacity at scale. A customer representing a large share of TSMC's most valuable business wields genuine negotiating power, co-plans capacity and roadmaps from a position of importance, and cannot be disappointed without serious consequence. The relationship is a mutual dependence between titans, not a supplier at the mercy of its customers or customers at the mercy of their supplier.
This mutual dependence mostly protects TSMC — the giants cannot get comparable chips elsewhere, so they stay and pay — but it caps the company's power and exposes it to concentration risk. TSMC cannot squeeze these customers the way a monopolist might squeeze fragmented buyers; it must keep them satisfied, invest to serve their roadmaps, and absorb their cyclicality. And the customers, motivated by both economics and the strategic desire for a second source, are perpetually willing to support any credible alternative that emerges. The customer lock-in is a genuine and formidable moat, but it is a bond between mutually dependent giants, and the mightiness of the buyers means TSMC's dominance on the demand side is real but not unlimited — it holds because there is no alternative, not because the customers lack the power to use one if it appeared.
- ReportedThe 20-F discloses reliance on a small number of major customers.TSMC Form 20-F, FY2025 (net revenue, noncurrent assets by country, customer concentration, capacity above 17 million wafers, risk factors; earthquake losses of about NT$3B in 2Q24 and NT$5.3B in 1Q25) — FY2025 (and prior years) · publ. Filed early 2026 · source ↗