TSMC has unveiled two new chip manufacturing nodes: the A13 for AI chips in 2029 and the N2U as a more cost-effective option for phones and laptops. For both processes, TSMC will continue to use existing ASML EUV machines rather than the new High-NA EUV variant, which costs twice as much.
TSMC unveiled two new chip manufacturing processes on Wednesday at the North America Technology Symposium 2026. The A13 node, a direct evolution of the A14 announced in 2025, is expected to enter production in 2029 and is primarily aimed at AI chips. The N2U is a more accessible variant intended for both AI and chips in phones and laptops, and is expected to enter production in 2028.
For both new processes, TSMC is therefore not opting for ASML’s so-called High-NA EUV machines. These cost approximately $400 million each, roughly twice as much as the current EUV generation.
TSMC has made a conscious decision to maximize the use of its existing machines. “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,” says Kevin Zhang, deputy co-chief operations officer and senior vice president at TSMC. “This is definitely a strength.”
Tip: TSMC prioritizes efficiency over raw performance
Chiplets are becoming the new driving force
The gains at the transistor level are modest, but TSMC expects the bigger performance leaps for AI chips to come from a different angle: combining multiple chips into a single package. Whereas Nvidia’s Vera Rubin, which is set to launch this year and manufactured by TSMC, consists of two compute chips and eight stacks of high-bandwidth memory, TSMC aims to combine ten large chips with twenty memory stacks by 2028.
Combining multiple chips presents its own challenges. Chips generate heat during use, and the various materials used in the packaging expand differently when heated. This adds a new layer of complexity for chip designers. At the same time, TSMC appeared earlier this year to be largely unaffected by the trade war and projected approximately 10 percent revenue growth by 2025, partly thanks to sustained demand for AI hardware.