You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
The adoption of ASML's High-NA EUV lithography tools is being delayed by major chipmakers due to their extremely high cost—about $360–400 million per machine—and uncertain return on investment. TSMC is taking a cautious approach, skipping High-NA for its upcoming 2nm and A16 nodes, instead extending the use of current EUV tools with multiple patterning. Intel, the first major customer to adopt High-NA, has faced significant financial losses in its foundry business, highlighting the risks of early adoption. Samsung has yet to commit to High-NA and may even wait for the next-generation Hyper-NA EUV, expected around 2030, due to reliability and cost concerns. While ASML projects mass production using High-NA EUV by 2026, most foundries are holding off until the technology proves more cost-effective and mature.
I see zero chance for Intel 14A to be HNA-EUV with HVM prior to 2030, as Pat G promised. Maybe they can run some HNA-EUV wafers with limited layers in OR since they already have 2 systems for R&D but not for mass production. Just my opinion of course.
ASML is supposedly shipping first EXE:5200 tool to Intel. This would be third high NA tool for Intel, if I didn't mix something up. If true, Intel must be serious about using high NA for 14a.
LBT’s perspective on 14A casts doubt on the future viability of both High NA and Hyper NA EUV.
Based on current trajectories, I estimate that TSMC may not adopt High NA EUV for HVM until around 2031, likely for the A10 node. Meanwhile, Intel and Samsung appear to be struggling to secure major customers beyond the 2nm node, raising questions about whether they can justify the investment in High NA EUV for HVM.
As for Hyper NA EUV, commercialization seems unlikely if TSMC remains the only potential customer—and even then, its adoption may not occur until around 2040.
LBT’s perspective on 14A casts doubt on the future viability of both High NA and Hyper NA EUV.
Based on current trajectories, I estimate that TSMC may not adopt High NA EUV for HVM until around 2031, likely for the A10 node. Meanwhile, Intel and Samsung appear to be struggling to secure major customers beyond the 2nm node, raising questions about whether they can justify the investment in High NA EUV for HVM.
As for Hyper NA EUV, commercialization seems unlikely if TSMC remains the only potential customer—and even then, its adoption may not occur until around 2040.
LBT’s perspective on 14A casts doubt on the future viability of both High NA and Hyper NA EUV.
Based on current trajectories, I estimate that TSMC may not adopt High NA EUV for HVM until around 2031, likely for the A10 node. Meanwhile, Intel and Samsung appear to be struggling to secure major customers beyond the 2nm node, raising questions about whether they can justify the investment in High NA EUV for HVM.
As for Hyper NA EUV, commercialization seems unlikely if TSMC remains the only potential customer—and even then, its adoption may not occur until around 2040.