• This is a political forum that is non-biased/non-partisan and treats every person's position on topics equally. This debate forum is not aligned to any political party. In today's politics, many ideas are split between and even within all the political parties. Often we find ourselves agreeing on one platform but some topics break our mold. We are here to discuss them in a civil political debate. If this is your first visit to our political forums, be sure to check out the RULES. Registering for debate politics is necessary before posting. Register today to participate - it's free!

Trump To Tariff Chips Made In Taiwan, Targeting TSMC

Tsmc is the only one getting good yields at the most advanced nodes.
AMD Zen 5 CPUs are made with TSMC's N4X node.

Apple M3 CPUs and Intel Arrow Lake CPUs are made with TSMC's N3B node.

Apple M4 CPUs are made with TSMC's N3E node.


Samsung has poor yields at anything below 4 nm, and Intel is a good generation behind Samsung to my understanding
Intel is having TSMC make their CPUs now.
 
I'm not going to pay it, either. I'll find other alternatives...or I just won't buy it. That means the manufacturer loses sales, which means they lose money.
Like Trump says, "They are going to come in because it's good for them to come in."
Which means they need to find a solution that doesn't mean they are affected by the tariffs.
Intel CPUs are made in Taiwan by TSMC.

Apple CPUs are made in Taiwan by TSMC.

AMD CPUs are made in Taiwan by TSMC.

Nvidia graphics cards are made in Taiwan by TSMC.

AMD graphics cards are made in Taiwan by TSMC.

What alternative are you envisioning here?
 
Intel CPUs are made in Taiwan by TSMC.

Apple CPUs are made in Taiwan by TSMC.

AMD CPUs are made in Taiwan by TSMC.

Nvidia graphics cards are made in Taiwan by TSMC.

AMD graphics cards are made in Taiwan by TSMC.

What alternative are you envisioning here?
Someone will have to make their chips somewhere other than Taiwan.
 
Someone will have to make their chips somewhere other than Taiwan.
How will they do that? TSMC is the company with the lithography machines to make them, and the only company that makes those lithography machines already has more than five years of backorders.
 
How will they do that? TSMC is the company with the lithography machines to make them, and the only company that makes those lithography machines already has more than five years of backorders.
shrug...

I don't know, but they'll have to figure out a way or they are going to lose a lot of sales.
 
shrug...
I don't know, but they'll have to figure out a way or they are going to lose a lot of sales.
I suppose that people in the United States could simply stop buying computers and smartphones.
 


Intel is in deep doodoo

It may abandon the plan to produce advanced computer chips. It either can't get customers or more likely its advanced foundries are not producing at yields it needs to be profitable.

Currently I am sitting at a 25% loss on my intel shares, so I don't like it.

That will leave TSMC at the top followed by Samsung which is having its own problems.

Intel has fallen dramatically in the last 10 years or so, it has never been able to get the EUV process right
 



MAGA voters, do you have ****ing idea how utterly ****ing stupid this is? Like totally mind-blowing stupid.

You voted for it.

I bet the Biden 450 billion dollar Chip and Science Act kept you awake at night.
 
Trump had a plan a few months ago, that I didn't think much of at the time, but maybe he was onto something.

His idea was for a company that would license TSMC manufacturing processes, make chips in the United States, and be co-owned by five different tech companies (TSMC, Apple, Nvidia, Broadcom, and Qualcomm).


Intel is in deep doodoo
Yes. Their last two revolutionary leaps forward in CPU design came in 2006 and 2011.

They had plans for a new revolutionary leap forward in CPU design, but then they could only afford to do either it or their new foundry processes, but not both, so they scrapped their work on the new CPU concepts and laid off the engineering team that was working on it to focus on their foundry processes.

I wonder, if they are now abandoning their foundry processes, will they try to resurrect their now-scrapped CPU design and rehire the engineers who were developing it?
 
Trump had a plan a few months ago, that I didn't think much of at the time, but maybe he was onto something.

His idea was for a company that would license TSMC manufacturing processes, make chips in the United States, and be co-owned by five different tech companies (TSMC, Apple, Nvidia, Broadcom, and Qualcomm).



Yes. Their last two revolutionary leaps forward in CPU design came in 2006 and 2011.

They had plans for a new revolutionary leap forward in CPU design, but then they could only afford to do either it or their new foundry processes, but not both, so they scrapped their work on the new CPU concepts and laid off the engineering team that was working on it to focus on their foundry processes.

I wonder, if they are now abandoning their foundry processes, will they try to resurrect their now-scrapped CPU design and rehire the engineers who were developing it?


Taiwan has restricted TSMC on what it can build in the US, so TSMC can't produce the most advanced nodes in the US
 
It will be better than nothing if Intel folds.


Intel will likely sell its foundries or spin it off like AMD did with global foundries.

When intel hit $ 30 I was expecting the government to step in and provide more support. Intel has fallen further and faster than I expected. I knew the plan to get to 1.4 nm was unrealistic in the time frame planned, but that they basically abandoned 2 nm, 1.8 nm and now potentially 1.4 nm, doesn't bode well
 
It will be better than nothing if Intel folds.
I can't see Intel folding. They have a habit of making comebacks.

I have a hard time grasping that they are working at the angstrom levels now. When I was in CMP, the 1 micron boundary was just broken. We got down to 0.12 microns when I left the industry.
 
I can't see Intel folding. They have a habit of making comebacks.

I have a hard time grasping that they are working at the angstrom levels now. When I was in CMP, the 1 micron boundary was just broken. We got down to 0.12 microns when I left the industry.
Economies of scale are not in intels favor

TSMC spends upwards of $20 billion on it most advanced foundries. It can spread that cost against multiple different customers to recoup the capital investment. Then when it is no longer cutting edge run for low value chips at discounted prices.

Intel really just has Intel as a customer. Until it can prove its nodes are reliable with good yield outside customers won't commit to using its foundries and potentially not get chips produced because they gave up space at TSMC to run at Intel
 
I have a hard time grasping that they are working at the angstrom levels now.
They aren't, actually. Those are marketing nanometers, not engineering nanometers.

What TSMC calls 3nm, Samsung calls 2nm, and Intel calls 18a, are actually the 7nm node in real world engineering measurements.
 
They aren't, actually. Those are marketing nanometers, not engineering nanometers.

What TSMC calls 3nm, Samsung calls 2nm, and Intel calls 18a, are actually the 7nm node in real world engineering measurements.


Which I think is likely the limit of how small the features can be considering atoms are 0.1 to 0.5 nm in size.
 
They aren't, actually. Those are marketing nanometers, not engineering nanometers.

What TSMC calls 3nm, Samsung calls 2nm, and Intel calls 18a, are actually the 7nm node in real world engineering measurements.
Any idea why? I wonder if the 18a is how flat they have to keep the chip for process resolution?

It's been more than 2 decades for me knowing the latest in the field. We went by line widths. Not nodes. Could the node size be the transistor region instead of the line widths?
 
Which I think is likely the limit of how small the features can be considering atoms are 0.1 to 0.5 nm in size.
They will be stuck on the 7nm node for a little while, but they will probably eventually achieve the 5nm node, probably when they start using High-NA Extreme Ultraviolet Lithography.

Probably the node that they will advertise as 1nm will be the 5nm node.
 
It's been more than 2 decades for me knowing the latest in the field. We went by line widths. Not nodes. Could the node size be the transistor region instead of the line widths?
They had to change transistor design so that it was no longer a two dimensional design. Below a certain size the two dimensional transistors stopped working. They had to switch to a 3D design. So they had to start estimating transistor size instead of directly measuring it.

Then TSMC started making wildly aggressive estimates for their transistor sizes to make it appear that they were more advanced than Intel when they weren't. Intel put up with it while their chips were actually better. But then Intel stagnated, and eventually TSMC surpassed them. And then Intel started making their own estimates just as aggressive as TSMC's estimates.
 
Taiwan has restricted TSMC on what it can build in the US, so TSMC can't produce the most advanced nodes in the US
I had another thought about that. We actually do not have too many nodes left to go before we can't go any smaller.

In less than ten years there probably aren't going to be any more advanced nodes for Taiwan to keep at home.
 
They had to change transistor design so that it was no longer a two dimensional design. Below a certain size the two dimensional transistors stopped working. They had to switch to a 3D design. So they had to start estimating transistor size instead of directly measuring it.
This isn't what I was speaking of. The transisters always took more width, and I believe we were doing 3D back then. I mean this was the that they could resolve to. It was used for the traces between transistors.

I was involved with CMP (Chemical Mechanical Planarization) and less fluent in the other areas.
 
This isn't what I was speaking of. The transisters always took more width, and I believe we were doing 3D back then. I mean this was the that they could resolve to. It was used for the traces between transistors.

This article covers some of the problems in naming current node sizes:
 
Back
Top Bottom