No more High N.A. EUV (ASML next gen lithography machine)? Not accurate. High N.A. will come, but slowly: 2028-29

We should really think about High NA as optionality value for ASML (share price), rather than a catalyst

• There’s been an avalanche of biased or disingenuous comments along the lines of “Samsung, TSMC won’t use High N.A. EUV” or “too expensive, too difficult”.

• Samsung will use High N.A. 4F² DRAM by 2028. TSMC will use High N.A. in some, but not all A14 processes by 2028. High NA direct cost is a partial issue, yields and density being as important in “total cost”.

• We should really think about High NA as optionality value for ASML (share price), rather than a catalyst.

Serious sources like The Bell in Korea says that Samsung won’t need High N.A. EUV for DRAM manufacturing; WCCF says that TSMC won’t use High N.A. for 14A. These are amplified by usually reliable semiwiki or TrendForce or eeNews.

All that is partially true but very exaggerated:

• Samsung won’t use High N.A. for 3D DRAM. Yes, the point of 3D DRAM is to use simpler lower density cell structure, but piled up vertically like NAND. But 3D DRAM is for now a concept that won’t exist in production before 2030-35. The next gen of DRAM, the 4F² cell structure will use High NA from 2028.

• TSMC won’t use High NA in A14 node. Not really, TSMC will use High NA in some but not all A14 process variations.

The issue with High NA EUV is 2 fold: