The highest estimate I found was 20bn.
If we are spotting intel 50% of the cost, why not go all the way and do it ourselves?
There are a lot of small european semi companies, both design and manufacturing, that we could this with instead. Why not give that money to Infineon?
Makes no sense.
Europe builds _1_ component for one fab stage (the EUV litho machines). For the other stages, there are other manufacturers in other countries.
Japan is still the dominate supplier of the chemicals used in silicon fabs.
The majority of the process is still DUV layers (less advance than EUV), and Japan's Nikon/Canon are the leaders in DUV.
https://seekingalpha.com/article/4412366-asml-not-dominant-w...
Secondly, TSMC and Samsung use EUV for only a few chip layers, and use DUV immersion for the rest, so both types of systems are needed at and below 7nm. The number of mask layers increases as nodes decrease.
A 28nm IC has up to 50 layers, a 14nm/10nm has 60 layers,
A 7nm IC has 80 layers, and
A 5nm IC has 100, depending on manufacturer.
TSMC uses EUV for just 12 layers at 7nm vs. 68 DUV layers. TSMC uses EUV for 22 layers at 5nm vs 78 DUV layers. Immersion DUV is critical to make these chips, and ASML is losing share to Nikon.Yeah, read the article. ASML market share in immersion DUV dropped from 95 to 85%.
We have our own semiconductor industry. Why not help them instead of intel?
Not that I support government subsidies but $10 billion would not build you a 5nm fab. Heck, it wouldn't even get you a 14nm, with the rate the Chinese are sucking up all the used fab equipment.