⚠ Switching Costs Fall If TSMC's Lead NarrowsModerate threat

The switching costs that lock in TSMC's customers rest on a crucial assumption: that switching means moving to an inferior process. That is what makes the trade self-defeating today. But the switching calculus changes the moment a rival offers a comparable or better process — then the customer is no longer trading down, and the redesign cost, while still real, is weighed against a competitive or superior alternative rather than an inferior one. The lock-in is therefore contingent on TSMC's technology lead: it is powerful precisely because, and only as long as, TSMC's process is the best available.

This ties the customer-lock-in moat to the technology moat beneath it. Should Intel or Samsung ever field a genuinely competitive leading-edge process, the switching costs that seem insurmountable today would look more like ordinary — high but surmountable — costs that a large customer might absorb for strategic resilience or a better deal, especially with geopolitical motivation to diversify. The redesign effort and ecosystem friction would remain, providing some stickiness, but the decisive factor — 'you'd be shipping a worse chip' — would fall away. So the customer lock-in, formidable as it is, is not independent of the process lead; it is largely an expression of it. TSMC's customers are locked in because leaving means losing, and that holds only while TSMC keeps winning the technology race. The moat within the moat is still, ultimately, the process lead — which is exactly what Intel's state-backed foundry drive is trying to close¹.