TAIPEI (Taiwan News) — TSMC said Wednesday it can produce smaller, faster chips using existing lithography tools, avoiding costly upgrades from ASML, Reuters reported.
The chipmaker, which supplies chips to Apple and Google, unveiled two new technologies: A13, set for production in 2029 and aimed at AI chips, and N2U, a lower-cost option for phones, laptops, and AI applications.
TSMC said A13, a shrink of A14, will reduce chip area by 6% and improve efficiency and performance, per Nikkei Asia. Generally, the smaller the process the better performance.
TSMC said it will continue to rely on its extreme-ultraviolet lithography machines from ASML, instead of shifting to newer “high numerical aperture” systems that cost about NT$12.63 billion (US$400 million). Executives said the strategy reflects strong research and development in extending the capabilities of existing tools.
“This is where I think our R&D has done exceptionally well in terms of leveraging existing EUV technology while setting an aggressive technology scaling roadmap,” TSMC Senior Vice President Kevin Zhang (張曉強) said. “This is definitely a strength.”
The company said performance gains from smaller and faster chips will be incremental. It is instead focusing on advanced packaging technologies that combine multiple chips, an area analysts say will drive future gains, especially for AI systems.
TSMC said it expects to be able to package up to 10 large chips and 20 memory stacks together by 2028, up from current designs such as Nvidia’s upcoming Vera Rubin platform. Analysts say this approach could help extend Moore’s Law by shifting from single-chip designs to multi-chip systems.
“Moore’s law is morphing from a monolithic, single die in a package to multi-die in a package,” said TechInsights Vice Chair Dan Hutcheson. “And that allows the power and performance gains.”
However, analysts said multi-chip designs present new engineering challenges, including heat management and material stress. Large chip packages can bend or crack, issues seen in Nvidia’s Rubin AI processor, according to Ian Cutress, chief analyst at consultancy More Than Moore.
