Excellent thread by @Gregory_C_Allen covering most key issues on trilateral export control negotiations. Only problem: January was just agreement "in principle" btwn Tokyo, The Hague, Washington. Hence short thread to clarify. ๐งต
I discussed major elements of the issue in this earlier thread:
Most the issues raised in this thread remain operative, despite reports that there is an "agreement".
Most the issues raised in this thread remain operative, despite reports that there is an "agreement".
When asked about details of the "agreement" this week at a House Foreign Affairs Committee hearing, he declined to provide a direct answer, noted that:
details โwould have to be discussed in a closed hearing.โ
foreignaffairs.house.gov
details โwould have to be discussed in a closed hearing.โ
foreignaffairs.house.gov
The reason a closed door would be needed is clear. The details are complicated, will likely require some compromise, + are not finalized yet. At some point though, details will need to be made public. Japan, for example, could issue regulations for public comment.
One of the major criticisms from the affected countries and companies was the unilateral nature of the 7 October controls, no comment period BEFORE they went into effect, and the messy process involved, ie, having to pull engineers out of Chinese facilities on short notice.
US allies like Japan and the Netherlands have previously preferred to work through more orderly and multilateral channels, like the Wassenaar Agreement, to discuss and implement new export controls.
The central issues remain: Neither the Dutch or Japanese government appears willing to mirror two critical elements of the 7 October controls:
-domestic persons controls
-restrictions specific to the production of memory semiconductors, DRAM and (3D) NAND
-domestic persons controls
-restrictions specific to the production of memory semiconductors, DRAM and (3D) NAND
Under 7 October controls, the Commerce Bureau of Industry and Security (BIS) took the unprecedented step of expanding an existing process to inform not just specific firms via confidential letter, but to inform via regulation all US persons of a new licensing obligation.
Specifically, BIS informed all U.S. persons, corporate and individual, that their activities directly or indirectly in support of the development or production of advanced node semiconductors in China require a license.
h/t Akin Gump akingump.com
h/t Akin Gump akingump.com
It seems unlikely that either the Japanese or Dutch government will implement this type of control. Traditionally they have only used persons controls in cases of weapons of mass destruction related activities, not clearly dual-use applications such as semiconductor tools.
Here for example, both the Japanese and Dutch response will likely mean nationals can continue to work at facilities to close out contractual arrangements, to avoid legal issues. Companies in both jurisdictions are concerned about blowback from Chinese authorities.
In addition, ASML almost certainly is pushing back on an agreement that would place all immersion lithography under controls. While this technology can be used for for certain layers @ 14/10/7 nm and <, this is stretching tech, which is not considered cutting edge in industry.
Advanced node manufacturing below 7 nm requires extreme ultraviolet (EUV) lithography, which is already restricted for export to China via provisions under the Wassenaar Agreement. ASML is barred from providing EUV systems to both Chinese firms and multinationals in China.
In China, there are already considerable numbers of immersion lithography systems. SMIC, China's leading foundry, for example, already has over 200 dense ultraviolet (DUV) lithography systems, the majority of them likely immersion lithography.
Hence, when details of a final agreement become known, it is unlikely to be close to a mirroring of 7 October controls, and could include nods to existing proposals before Wassenaar such as GAAFET-related controls. The devil is in the details here and it is devilishly complex.
And just to clarify, these types of tools can be used in many ways and are not designed to be specific for a particular technology node. The complex reasons why a particular tool would be used, issues like commercial yields, costs, etc are outside the scope of this thread.
Loading suggestions...