⚠ Big Customers Can Dual-Source DeliberatelyModerate threat

Deep co-development binds customers tightly, but the largest and most sophisticated of them have both the resources and the motive to resist total lock-in by qualifying a second source. Some designers deliberately maintain relationships with more than one foundry, or design certain products to be more portable, precisely to preserve negotiating leverage and supply resilience — and the strategic imperative to reduce dependence on a single Taiwan-based supplier gives them, and their governments, extra reason to do so. A giant like Apple or Nvidia, spending billions with TSMC, has the scale to fund a second-source effort if it chose to, and the AI hyperscalers designing their own chips have every incentive to keep their options open.

In practice, dual-sourcing at the leading edge is severely constrained by the simple fact that there is no comparable second source — Samsung and Intel are a generation or more behind, so qualifying them means accepting an inferior chip. That is why, despite the incentives, the biggest customers overwhelmingly rely on TSMC for their most advanced products. But the desire for a second source is real and persistent, backed by strategic and geopolitical motivation, and it means TSMC's largest customers are actively willing to support and qualify alternatives the moment one becomes credible. The co-development lock-in is powerful, but the most powerful customers are working to keep it from becoming absolute — and a credible rival would find willing partners in TSMC's own customer base — Samsung remains the one other leading-edge option, and customers have used it when terms suited¹.