TSMC SVP Kevin Zhang says ASML's new machines, which cost €350M each, are too expensive, and the A16 node, due in late 2026, won't need to use the new machines
“The cost is very high,” TSMC Senior Vice President Kevin Zhang said at a technology symposium in Amsterdam on Tuesday …
Context & Ripple Effects
TSMC’s A16 decision put the economics of high-NA EUV at the center of leading-edge manufacturing planning, rather than treating the newest lithography gear as an automatic node requirement. Weeks later, ASML said TSMC, Intel and Samsung would receive the systems, making deployment—not access—the key question.
The initial caution foreshadowed TSMC’s later decision to defer high-NA EUV in production through 2029. It also sits within a broader supplier–foundry pricing conflict, as later reports said TSMC was resisting ASML’s proposed equipment price increases.
First-order effects
- TSMC can pursue A16 without buying or qualifying ASML’s €350M high-NA EUV tools for that node, avoiding that immediate capital outlay.
- ASML loses an early production-use case at its largest foundry customer, even as the machines remain part of its customer delivery plans.
Second-order effects
- High-NA EUV’s commercial case shifts toward proving a sufficiently large yield, density or operating-cost advantage over existing tools; tool deliveries alone do not ensure volume deployment.
- Intel and Samsung gain a clearer competitive choice: deploy high-NA earlier to differentiate, or preserve capital if TSMC’s conventional-tool approach remains viable.
Third-order effects
- Leading-edge chip progress may become more selective about adopting each new manufacturing generation, with foundries timing equipment transitions around total production economics rather than node labels alone.
- If high-NA adoption remains delayed across major foundries, ASML’s revenue mix and customer negotiations could be shaped more by upgrade timing and pricing discipline than by immediate broad rollout.
The trend: Advanced-chip manufacturing is moving toward a tighter compute-economics discipline, where even indispensable equipment suppliers must demonstrate a near-term return before foundries scale the newest tools.