US give TSMC three options

[tonyget]

TSMC’s first board meeting in the United States on the 12th did not announce further plant expansion plans in the United States or personnel cases such as Wang Yinglang and Zhang Zongsheng as expected.

However, it is worth noting that the market is paying close attention to the subsequent off-the-table discussions between the US government and TSMC. It is rumored that the United States may give TSMC "three paths" to choose from, in order to achieve the ambitious goal of "Made in the United States" and at the same time save Intel (Intel).

It is understood that there are estimated to be three plans proposed by the United States:

First option, TSMC directly builds advanced packaging plants in the United States to complete one-stop services.

Regarding the first point, in fact, TSMC has not been very willing to build packaging plants in the United States. Mainly due to the shortage of manpower and poor gross profit margin, the simultaneous construction of packaging plants may also be detrimental to local packaging plants such as Amkor and Intel.

In October 2024, TSMC and Amkor have signed a memorandum of cooperation to provide advanced packaging and testing services in Arizona. Close cooperation between the two parties can shorten the overall product production cycle.

Intel's advanced packaging base is not small either. It has previously announced that it will expand investment in its advanced packaging plant in New Mexico and accelerate the development of 3D packaging technology Foveros.

Second option, the U.S. government and many major manufacturers such as TSMC have jointly invested in Intel's independent foundry business, including TSMC's concurrent technology investment.

The second option is for the U.S. government to propose a joint venture plan, hoping that TSMC and a number of major manufacturers will "jointly participate in this grand initiative" and jointly invest in Intel's wafer foundry business, including TSMC providing technology transfer.

Third option, Intel, which is as advanced as TSMC in advanced packaging, will directly undertake subsequent packaging orders from TSMC’s customers in the United States.

The third option is for Intel to directly undertake subsequent packaging orders from TSMC’s customers in the United States. For example, Apple, which is determined to produce films at TSMC's U.S. factory, is quite familiar with cooperating with Intel.

Semiconductor industry players said that judging from the fact that the U.S. government is taking shortcuts to strengthen local "American manufacturing" and will try its best to keep Intel alive, TSMC is almost the only hope for rescue.

台積赴美董事會檯面下祕議　傳川普拋「三條路」施壓

台積電12日首度赴美召開董事會的決議，並未如外界猜測，宣布美國進一步的擴廠計畫，或是王英郎、張宗生等人事案。

www.digitimes.com.tw

 

[Daniel Nenni]

This begs the question: Is the US preparing for China to take over Taiwan?

First option, TSMC directly builds advanced packaging plants in the United States to complete one-stop services.

This would be easy enough to do, much easier than building a fab in the US.


Second option, the U.S. government and many major manufacturers such as TSMC have jointly invested in Intel's independent foundry business, including TSMC's concurrent technology investment.

I do not see how this would work. The big appeal of Intel Foundry is to have a second source for leading edge semiconductor manufacturing. Competition keeps semiconductors competitive in all phases of design and manufacture.


Third option, Intel, which is as advanced as TSMC in advanced packaging, will directly undertake subsequent packaging orders from TSMC’s customers in the United States.

I do not see this happening. Intel and TSMC packaging are not the same, TSMC is in fact far ahead of Intel in customer packaging. Also, packaging is a competitive moat for TSMC which I do not see them giving up without a fight. Seriously, packaging has been a huge investment for TSMC and they are absolutely winning.

 

[Mooredaddy]

I hate all of this from a tsmc investor standpoint. Why is the U.S gov trying to strongarm tsmc when they have already gone over and above building fabs in Arizona.

 

[soAsian]

I hate all of this from a tsmc investor standpoint. Why is the U.S gov trying to strongarm tsmc when they have already gone over and above building fabs in Arizona.

The Trump administration = transaction based. It's not like Biden is any better, as he mostly kept what Trump did with tariffs and maintained most of Trump's business framework. It looks like America is becoming more China-like in business.

 

[Paul2]

I think this is a pre-Trump policy echo, and now they will get something new completely again.

 

[Arthur Hanson]

Control of advanced chip manufacturing is key to a modern, leading economy and the US needs TSM to stay on top, the only question is how to cultivate a high-tech economy and retain economic leadership. So far, the US the US and allies are the leaders and hope to stay that way.

 

[blueone]

I hate all of this from a tsmc investor standpoint. Why is the U.S gov trying to strongarm tsmc when they have already gone over and above building fabs in Arizona.

Because even the government dummies must know fabs are not enough. The problem is that they are still too ignorant to understand what fab process development really means (which will stay in Taiwan no matter what happens), and that this isn't like other joint ventures, even jet engines. (CFM is a joint venture.) Chip manufacturing is uniquely difficult, uniquely expensive, and apparently far too technical for politicians to comprehend. Even AI technology is more comprehensible. Only quantum computing is more obscure, IMO.

 

[MKWVentures]

Lets ask some questions:

1) Advanced packaging means CoWoS like? or 4000 pin BGA/LGA? or 6 die Logic+DRAM stack? The first is responsible for about 1% of the units packaged in the world

2) The reason fabs were not built in US was Capex subsidies. That is 100% fixed. The reason packaging (non CoWoS) is due to labor cost, facility cost, and enormous scale needed. Packaging in the US is >2x the cost of other countries from the exact same company. So how can you have a cost effective packaging site in the US?

3) Even then, 90% electronic systems would be made in Asia.... motherboards, Phones, PCs, Servers are made in Asia. So what is the plan here? Again issue is labor and scale. Chinese/Taiwanese/Malaysian factories are 10x the scale of the factories in the US.

Fabs are great in the US and always make sense. It is not clear how packaging (other than CoWoS) or systems are possible in the US.

None of this has anything to do with Intel IMO. Intel's problem is that is is not competitive with TSMC regardless of location due to cost and scale and finances around that. What is the plan to make Intel competitive? Hurt TSMC as much as possible? Ban/tariff the most successful US companies from using TSMC? what is the plan?

The answer IMO is let the best company win and the government can provide credits when desired to achieve a result. Incentives to build in certain areas (other than Arizona, California, NY) would be great. Reminder: never give the money upfront.... its always a rebate after purchase.

 

[BruceA]

TSMC and Taiwan should just make a deal with PRC.

All US wealth and leadership in technology starts and ends with TSMC.

If TSMC was to be controlled by PRC and allowed to again sell to Huawei and ship advanced node to other companies in PRC we’d have a real horse race. Of course Us could respond by embargo and the whole world go into depression, who blinks first ?

 

[blueone]

TSMC and Taiwan should just make a deal with PRC.

All US wealth and leadership in technology starts and ends with TSMC.

If TSMC was to be controlled by PRC and allowed to again sell to Huawei and ship advanced node to other companies in PRC we’d have a real horse race. Of course Us could respond by embargo and the whole world go into depression, who blinks first ?

This is my nightmare scenario I've posted here more than once. Fortunately, everyone who has responded thinks this is impossible, partially due to the way the PRC has handled Hong Kong, so the Taiwanese will never agree. I'm glad no one agrees, because it helps me sleep at night.

 

[ai268]

Maybe just maybe, it's all about INTC stock price. There are at least three hedge fund managers, who have many wall street buddies, inside the Trump's cabinet. Trump's TikTok reversal is another example. Do billionaire give a damn about the lower 99.9999...% population?

 

[nghanayem]

Lets ask some questions:

1) Advanced packaging means CoWoS like? or 4000 pin BGA/LGA? or 6 die Logic+DRAM stack? The first is responsible for about 1% of the units packaged in the world

Depends on who you ask. If you ask a fab operator they say advanced packaging means 2.5D (COWOS, InFLO, EMIB, Foveros with passive base die, SiliconBox, etc) and 3D stacking (memory cubes (HBM), Foveros with active base die, SoIC, X-cube, etc.). I suppose you could also call CMOS image sensors, BSPDNs, and bonded separate periphery CMOS to memory array a 3D die stack too. But I don't think I have ever come across someone with that opinion, and most people will just leave it as similar but different technologies.

If you ask an OSAT... Almost everything more complex than wire bonding and leadframe is called advanced packaging. System in package, wafer level packaging, PoP, flip chip/BGA, top metal/RDL layers, wafer bumping/C4, 2.5D, 3D, etc.

2) The reason fabs were not built in US was Capex subsidies. That is 100% fixed. The reason packaging (non CoWoS) is due to labor cost, facility cost, and enormous scale needed. Packaging in the US is >2x the cost of other countries from the exact same company. So how can you have a cost effective packaging site in the US?

The cost delta is nowhere near that high, facility cost isn't that high, and modern assembly test is nowhere near manual as it used to be or people imagine it is. Modern OSAT facilities are fairly automated (not as much as a fab but not rows of people hand soldering things under a magnifying glass either). The biggest problem is the margins are abysmal around 10%, with what OSATs term traditional assembly test having low single digit margins. Margins that low mean you can barely afford to spend money on capex and things like advanced packaging are all in levels of investment. These abysmal margins also make the labor costs critical. Even if labor is only a 5% cost adder. If you are doing basic test or wire bonding that puts you in the red. So cost of labor is important but not because of why you think. I wonder why Taiwan has so many packaging sites when labor is more expensive than PRC/ROK, and much more expensive than Malaysia or Vietnam. Probably just a scale and existing investment over coming the labor cost thing.

As for packaging not being done in the US. TI (development/manufacturing in Texas and discrete transistor packages scattered about the east coast), Micron (for sure probe at all US sites plus if memory serves from their 2020 expansion a bit of regular assembly test for their legacy memory fab in VA), Intel (wafer prep/bumping and advanced packaging), Microchip CO/OR, OnSemi OR, Renasis (FL), and GF NY (3DIC, a little tiny bit of a bumping/wafer sort, and maybe photonics packaging). You also have some smaller fries like Grumman's foundry division, sky water TDK, NHanced, plus like a few others I don't remember.

While it isn't US Micron has traditional assembly test in Singapore which is far from low cost of living, and AMKOR has a bumping facility in Portugal that also does some regular assembly test.

In the first half of last year Intel said their advanced packaging and their traditional assembly test businesses broke even during like the Q2 earnings. Back during the pandemic Micron expanded their Virginia fab and increased back end capability/capacity to better support their long lifecycle product customers. You also have SK group opening a subsidiary back in like 2020 for glass substrate panel production and development in Georgia, and last year SK-Hynix chose to build a 3DIC fab in Indiana. Considering TSMC'S advanced packaging margin sits in that 30-40% range, and the above information I don't see why an advanced packaging fab can't be profitable in the US. Especially since advanced packaging and bumping is more fab-like than traditional OSAT-like.

 

[blueone]

Maybe just maybe, it's all about INTC stock price. There are at least three hedge fund managers, who have many wall street buddies, inside the Trump's cabinet. Trump's TikTok reversal is another example. Do billionaire give a damn about the lower 99.9999...% population?

Unlikely that this is just about stock price manipulation. If discovered and prosecuted it would mean years of prison time for anyone involved. Don't think so? Ask Bernie Madoff. Oh, you can't, he died in federal prison.

As for whether or not billionaires care about people who aren't billionaires, logic says that they do. They wouldn't be billionaires without many tens of millions of people investing their retirement accounts in the stock markets. Are billionaires altruistic at the level so-called progressives in the US would like them to be? Many are not, but some are, like Bill Gates, who feel guilty about their wealth and get serious about philanthropy . When Gates and other super-rich people lament that they aren't taxed enough though, I find it nauseating. Dear Bill and friends - if you really believe that, put your money where your mouth is and write the US Treasury a check. They accept donations. Challenge your fellow guilty-feeling ultra-billionaires to step out of their personal Gulfstreams on the way to Davos to do the same. For your convenience, here's the address, Bill:

Gifts to the United States
U.S. Department of the Treasury
Reporting and Analysis Branch 2
P.O. Box 1328
Parkersburg, WV 26106-1328

 

[vslee]

Dear Bill and friends - if you really believe that, put your money where your mouth is and write the US Treasury a check. They accept donations. Challenge your fellow guilty-feeling ultra-billionaires to step out of their personal Gulfstreams on the way to Davos to do the same.

It doesn't work if only he (or a small group) does it. Even a few multiples of his entire net worth would not make a dent in the national debt. It only makes a difference if there is a tax policy that covers all of the rich. Not just a few volunteers.

 

[blueone]

It doesn't work if only he (or a small group) does it. Even a few multiples of his entire net worth would not make a dent in the national debt. It only makes a difference if there is a tax policy that covers all of the rich. Not just a few volunteers.

I agree, but that wasn't my objective. Even if we confiscated the entire net worth of the top ten richest people in the US, you'd only get about $1.5T. And that's just a fantasy anyway, because much of their wealth invested in the stock markets, and if those equities were sold off it would depress the prices. The annual USG budget deficit is greater than $1.8T, so any talk of reducing the national debt this way is also a fantasy. My objective was just to have Gates put up or shut up, a guy who endowed his foundation with tens of billions of dollars, where it isn't taxed, and donated an additional $350M to offset part of the $500M cost of the foundation's fancy headquarters building in Seattle.

 

[vslee]

donated an additional $350M to offset part of the $500M cost of the foundation's fancy headquarters building in Seattle.

That is clearly excessive.

I do wish more of the ultrawealthy would either stop lobbying for tax cuts for themselves, if not advocate for a more progressive tax structure (as Gates is doing). But like you said, there are many loopholes for them to take advantage of anyways.

 

[blueone]

I do wish more of the ultrawealthy would either stop lobbying for tax cuts for themselves, if not advocate for a more progressive tax structure (as Gates is doing). But like you said, there are many loopholes for them to take advantage of anyways.

Actually, the US tax system is already highly progressive. The top 1% of income earners pay over 40% of federal income taxes, even with the silly "loopholes" Congress sneaks in.

This ranting was in response to the notion that the run-up in INTC might have been due to billionaires playing stock manipulation games. There's no evidence to support that conjecture. More likely this is due to rumors and day-trader silliness fed by posters on Reddit forums, who lie in wait for rumors. And the press nowadays has no pride when it comes to spreading rumors. INTC as a meme stock - who would have predicted that?

 

[ai268]

Buy the Rumor, Sell the News. But that's NOT what hedge does. Who were spreading the rumor? INTC is Back to $19 handle from around $28 a month ago.

 

[lefty]

This begs the question: Is the US preparing for China to take over Taiwan?

First option, TSMC directly builds advanced packaging plants in the United States to complete one-stop services.

This would be easy enough to do, much easier than building a fab in the US.


Second option, the U.S. government and many major manufacturers such as TSMC have jointly invested in Intel's independent foundry business, including TSMC's concurrent technology investment.

I do not see how this would work. The big appeal of Intel Foundry is to have a second source for leading edge semiconductor manufacturing. Competition keeps semiconductors competitive in all phases of design and manufacture.


Third option, Intel, which is as advanced as TSMC in advanced packaging, will directly undertake subsequent packaging orders from TSMC’s customers in the United States.

I do not see this happening. Intel and TSMC packaging are not the same, TSMC is in fact far ahead of Intel in customer packaging. Also, packaging is a competitive moat for TSMC which I do not see them giving up without a fight. Seriously, packaging has been a huge investment for TSMC and they are absolutely winning.

There's also a 4th option: no packaging in US.
TSMC has already promised 100$ billion investment - that should be enough.

 

[Daniel Nenni]

TSMC announced packaging in the US as part of their $100M in vestment, so it is option #1.