3 ms·

The big question in semiconductors is what will happen to the China-Taiwan situation. China has repeatedly stated that they intend to reunify with Taiwan by 204

by lacker 4y ago

The big question in semiconductors is what will happen to the China-Taiwan situation. China has repeatedly stated that they intend to reunify with Taiwan by 2049 for the 100th anniversary of their communist party. I think they will probably try. If there's fighting then there's a good chance that TSMC ends up destroyed. Refugees can flee, physical plants can be destroyed. So if TSMC is destroyed, what happens to everyone downstream who depends on semiconductors? We probably end up both delaying products for years and giving Intel boatloads of money thinking "I sure hope they can scale up their only-three-years-behind technology quickly." So we should subsidize Intel to keep improving their currently-uncompetitive technology. This isn't the same as most "we should make things in America" complaints, it's a contingency plan for a specific, reasonably likely intense disruption to the world of technology.

User23 4y ago
> This isn't the same as most "we should make things in America" complaints How is it any different? There is a strategic argument for onshoring pretty much anything that isn’t utterly frivolous like fidget spinners and such. Look how effectively Europe’s need for energy has undermined the Russia sanctions. The Ruble is stronger than it was before the invasion! It’s nice not to be vulnerable to foreign embargoes of food or energy, just to name a couple important strategically important things besides semiconductors.
- bsder 4y ago
  > There is a strategic argument for onshoring pretty much anything that isn’t utterly frivolous I would probably adjust that to "onshore anything that takes longer than <insert number of years> to spin up". You can spin up funiture factories in a year or less. You probably don't need them onshored. A steel mill probably takes 3-5 years. You probably should have a local alternative. You can't spin up ASML in less than 10 years. There definitely should be a local ASML alternative.
- mywittyname 4y ago
  > anything that isn’t utterly frivolous like fidget spinners and such. This frivolity is exactly what countries need to do if they want to manufacture at the level China is. Fidget spinners were built as a way of scaling up the manufacturing of very high-tolerance bearings, and selling them in the USA for $5 helped ensure that every American involved in manufacturing understood just how adept China had become.
ant6n 4y ago
> We probably end up both delaying products for years and giving Intel boatloads of money thinking "I sure hope they can scale up their only-three-years-behind technology quickly." It’s weird, five years ago they were two years ahead of everybody.
bagacrap 4y ago
Isn't tsmc building fabs in Arizona and Japan? I don't think China will invade Taiwan but what do I know. Is it really worth the risk of being cut off from international trade and banking the way Russia has been? If China invades and there's an international embargo of Chinese goods, semiconductors are but one small part of the havoc that would ensue.
- ksec 4y ago
  >Isn't tsmc building fabs in Arizona and Japan? Both planned Fabs, and projected growth in Fab capacity in both location combined wouldn't equal to a single MegaFab in TSMC Taiwan.

by lacker 4y ago

User23 4y ago

> This isn't the same as most "we should make things in America" complaints How is it any different? There is a strategic argument for onshoring pretty much anything that isn’t utterly frivolous like fidget spinners and such. Look how effectively Europe’s need for energy has undermined the Russia sanctions. The Ruble is stronger than it was before the invasion! It’s nice not to be vulnerable to foreign embargoes of food or energy, just to name a couple important strategically important things besides semiconductors.

bsder 4y ago
> There is a strategic argument for onshoring pretty much anything that isn’t utterly frivolous I would probably adjust that to "onshore anything that takes longer than <insert number of years> to spin up". You can spin up funiture factories in a year or less. You probably don't need them onshored. A steel mill probably takes 3-5 years. You probably should have a local alternative. You can't spin up ASML in less than 10 years. There definitely should be a local ASML alternative.
mywittyname 4y ago
> anything that isn’t utterly frivolous like fidget spinners and such. This frivolity is exactly what countries need to do if they want to manufacture at the level China is. Fidget spinners were built as a way of scaling up the manufacturing of very high-tolerance bearings, and selling them in the USA for $5 helped ensure that every American involved in manufacturing understood just how adept China had become.

ant6n 4y ago

> We probably end up both delaying products for years and giving Intel boatloads of money thinking "I sure hope they can scale up their only-three-years-behind technology quickly." It’s weird, five years ago they were two years ahead of everybody.

bagacrap 4y ago

Isn't tsmc building fabs in Arizona and Japan? I don't think China will invade Taiwan but what do I know. Is it really worth the risk of being cut off from international trade and banking the way Russia has been? If China invades and there's an international embargo of Chinese goods, semiconductors are but one small part of the havoc that would ensue.

ksec 4y ago
>Isn't tsmc building fabs in Arizona and Japan? Both planned Fabs, and projected growth in Fab capacity in both location combined wouldn't equal to a single MegaFab in TSMC Taiwan.