The Fall of Intel

[dada_dave]

I don't know much about fabs, other than ES2 (European Silicon Structures) from long ago. But I suspect sometime in the future Taiwan will not exist as country, and will become another Chinese province. With TSMC becoming a Chinese company and having a different name.

As a result, it seems China would have a ton of leverage over US companies and their tech products, in both commercial and defense/aerospace/military spaces, having to rely on Chinese fabs (including TSMC's Arizona fab). Thus I can see the motivation for getting Intel and perhaps other US fabs (Texas Instruments) up to snuff.

The above is just me spitballing. Thoughts?

Most plausible circumstances of Taiwan becoming non-independent anytime in the foreseeable future are the result of the use of force. Now there are degrees in the use of force beyond just outright invasion but even so. While it’s true that diversity in chip manufacturing would be important, so many other things would be going wrong at that point … that isn’t to suggest China wouldn’t try, but computer companies would have much bigger problems than just chips. The entire supply chain, even parts not directly related to China collapse.

 

[mr_roboto]

having done something like this (two different processes, albeit our own fabs), what ends up happening, probably, is you create your own “least common denominator” design rules that work with both fabs. But you inevitable give something up in doing this. For example, fab A might say minimum polygon area is 1nm^2 and minimum M0 spacing is 0,5nm, and fab B might say minimum polygon area is 1.2nm^2 and minimum M0 spacing is 0.3nm. So you end up with 1.2nm^2 and 0.5nm. Worst of both worlds.

And that’s assuming both processes give you equivalent number of metal layers and that the layer thicknesses are the same, dielectrics are the same, etc. If the layer thicknesses and dielectrics vary, then you have additional problems. You end up over engineering to make cycle time and satisfy hold times on both processes, and the design process takes a lot longer.

These days it seems like there's fundamental incompatibilities that should completely prevent using the same artwork with both fabs. AFAIK nobody does FinFETs exactly the same way, same goes for incoming new technologies like GAA and backside power delivery.

 

[Citysnaps]

Most plausible circumstances of Taiwan becoming non-independent anytime in the foreseeable future are the result of the use of force. Now there are degrees in the use of force beyond just outright invasion but even so. While it’s true that diversity in chip manufacturing would be important, so many other things would be going wrong at that point … that isn’t to suggest China wouldn’t try, but computer companies would have much bigger problems than just chips. The entire supply chain, even parts not directly related to China collapse.

All good points. At some time in the future China will likely make a move on Taiwan. Could be ten/twenty or more years. No doubt they have gamed out different scenarios and potential consequences. Should that happen I could see China keeping their best semiconductor processes/fabrication for only in-country use. Especially for aerospace/military applications.

It seems that would be a good time for the US to invest heavily and take more control over its in-country semiconductor fabrication going into the future. And reducing the possibility of China exerting leverage over the US.

 

[Cmaier]

These days it seems like there's fundamental incompatibilities that should completely prevent using the same artwork with both fabs. AFAIK nobody does FinFETs exactly the same way, same goes for incoming new technologies like GAA and backside power delivery.

That’s comparatively easy to cope with, especially since these are standard cell designs. You just have two sets of standard cells. The problem is that the size of a given cell (i.e. an INx4) is going to be bigger than it needs to be, because it will be the size of whichever inverter is bigger. And since Apple probably relies mostly on vendor-supplied cells (or so I was told by an apple engineer once), this would require them to do it themselves.

RAM blocks and PLLs are a problem, of course… No choice but to do that work twice.

They will also run into issues where they may have to wall off engineers - those who receive TSMC design rules and technical data may not be allowed to work with Intel, and vice versa.

In short, it’s all a bunch of extra work and will result in a worse chip. Unless they simply redo the entire physical design twice (in which case it’s even more extra work, but the chip won’t be worse than it could be)

 

[Cmaier]

It seems that would be a good time for the US to invest heavily and take more control over its in-country semiconductor fabrication going into the future. And reducing the possibility of China exerting leverage over the US.

I don’t want the government taking control of semiconductor fabrication. They’re not good at it. The people in charge of the government don’t understand it.

I am, however, all in favor of them subsidizing it, tariffing chip imports, demanding that government purchases have US-fabbed chips, etc. But before they can do some of those things, there must be a US capability to make bleeding edge chips, so they would need to do these things in stages.

 

[Citysnaps]

I don’t want the government taking control of semiconductor fabrication. They’re not good at it. The people in charge of the government don’t understand it.

I am, however, all in favor of them subsidizing it, tariffing chip imports, demanding that government purchases have US-fabbed chips, etc. But before they can do some of those things, there must be a US capability to make bleeding edge chips, so they would need to do these things in stages.

Yes... the government not making fabrication decisions, rather just their funding. Perhaps a small committee of esteemed technologists/scientists/etc. would make the decisions.

 

[Cmaier]

Yes... the government not making fabrication decisions, rather just their funding. Perhaps a small committee of esteemed technologists/scientists/etc. would make the decisions.

The only decisions the gov’t should make is who to fund.

 

[dada_dave]

All good points. At some time in the future China will likely make a move on Taiwan. Could be ten/twenty or more years. No doubt they have gamed out different scenarios and potential consequences. Should that happen I could see China keeping their best semiconductor processes/fabrication for only in-country use. Especially for aerospace/military applications.

Most Taiwanese resistance scenarios ensure that the Chinese would get bupkis in the event of a forced integration, that it would take decades to rebuild, and any talent needed to do so would be long gone and out of reach. The Taiwanese are very open about this and the measures they plan to take. Of course the Chinese know this which is why, were they to invade anyway, they would do so knowing that they would likely not be gaining TSMC in any meaningful capacity.

It seems that would be a good time for the US to invest heavily and take more control over its in-country semiconductor fabrication going into the future. And reducing the possibility of China exerting leverage over the US.

That is the thinking, yes. Though again more that TSMC would cease to exist as a going concern.

 

[Yoused]

Meanwhile, Samsung has gradually improved their GAAFET process, reaching yields that are poor rather than abysmal. If they continue to improve their "2nm", they could conceivably be a viable alternative to a swallowed-up TSMC.

OTOH, TSMC has been venturing into developing and burning SOT-MRAMs. This I think is likely to be the next advance in computer design, assuming it can be packaged with sufficient shielding to protect from environmental EM. There are probably other companies exploring this, but that TSMC is in the hunt makes me optimistic.

 

[cbum]

The only decisions the gov’t should make is who to fund.

Yes, except that right now, that would fund the one(s) who grovel the hardest.

 

[Yoused]

The only decisions the gov’t should make is who to fund.

It would be interesting to see an open GSE program, where a business could apply for government sponsorship and, if accepted, would receive appropriate subsidies, cut rate loans and certain other advantages (facilitated market access) but would first have to be vetted in public hearings and would have to assiduously adhere to labor and standards regulations.

 

[Nycturne]

It would be interesting to see an open GSE program, where a business could apply for government sponsorship and, if accepted, would receive appropriate subsidies, cut rate loans and certain other advantages (facilitated market access) but would first have to be vetted in public hearings and would have to assiduously adhere to labor and standards regulations.

This is generally the problem I have with these sort of subsidies. If the public is going to fund a thing, the public should be getting something specific from it, IMO. I'm not against that thing being "send a probe to Mars to find out if there's life". Knowledge brought into the public square is getting something out of it. What I'm less enthusiastic about is what we've seen are the games around things like fiber, where ISPs get incentivized to change the requirements after they've got the money, lay fiber past the users it is meant to serve without them ever getting service, etc. And with the current political climate, there's definitely a push to remove many of these requirements which makes it even easier to get these public funds and then not provide the thing those funds are supposed to be helping make real. Turning it more into a slush fund that big business can dip into. The whole "Privatize profit, socialize cost" nonsense.

And I am just not confident that if we were to fund local fabs for national security reasons/etc, that it wouldn't wind up being one of these slush funds.

 

[cbum]

And I am just not confident that if we were to fund local fabs for national security reasons/etc, that it wouldn't wind up being one of these slush funds.

Agree, but we may not have a choice or be left behind. The only solution would be to "throw the bums out" and install responsible representatives...
One can always dream.

 

[Nycturne]

Agree, but we may not have a choice or be left behind.

This is where my cynicism shows. In this climate, I would argue there's a third possibility: We fund things, and still fall behind. If it becomes a slush fund, what stops a company like Intel from taking the money, but still leaving us with just TSMC anyways? This is why I think the requirements and the ability to enforce them are so critical. Because if you have no teeth to hold people accountable to the agreements, you aren't going to get what you want, funding or no.

 

[Yoused]

If it becomes a slush fund, what stops a company like Intel from taking the money, but still leaving us with just TSMC anyways?

Then, I would attach very strict requirements to it: you do this thing, if you fail to do this thing, your company loses its charter, its IP and if you cannot account for where those funds went, we recover them from you personally. These executroids do not understand real consequences. It is up to us to lesson them.

 

[cbum]

I would attach very strict requirements to it

Yes, and the orange KingPIN will just ignore all that...

 

[Cmaier]

Yes, and the orange KingPIN will just ignore all that...

BZZT! Politics warning…

Let’s stick to theories of government intervention without triggering anyone.

 

[Nycturne]

Then, I would attach very strict requirements to it: you do this thing, if you fail to do this thing, your company loses its charter, its IP and if you cannot account for where those funds went, we recover them from you personally. These executroids do not understand real consequences. It is up to us to lesson them.

You do know that’s one of the key points in my posts on this topic? And I mention that in the very sentence after the bit you quoted?

 

[Cmaier]

it seems like a lot of this discussion is moot. If the U.S. wants an indigenous alternative to TSMC, the choices are slim. Once you get past Intel, who else is there? GlobalFoundries?

 

[casperes1996]

it seems like a lot of this discussion is moot. If the U.S. wants an indigenous alternative to TSMC, the choices are slim. Once you get past Intel, who else is there? GlobalFoundries?

I had no idea GloFo was from the United States. I always assumed it was Asian.