News TSMC and Intel foundry joint venture reportedly still in the works — AMD, Broadcom, and Nvidia approached

I still fail to see the logic behind this or why any of the governments who'd need to sign off on it would.

TSMC already has around 67% of all foundry revenue in Q4 (Samsung is second at ~8%) according to TrendForce and that list doesn't include Intel. According to Intel's Q4 2024 financials foundry revenue was $4.5b which is higher than everyone other than TSMC and makes up around 17% of TSMC's revenue for Q4. Just adding Intel in as is (no big contract work yet) would push TSMC a bit over 70% while dropping Samsung to ~7.5%.

Then there's the practical side of this venture requiring guaranteed wafer buys from Intel which as we all know worked out real well for AMD with GloFo.

All this being said given the sheer incompetence of the Intel board and its idiotic makeup I wouldn't put it past them to try.
 
TSMC would love to have a monopoly on advanced nodes. What a terrible idea, just as Intel foundry services is ramping their leading edge 18A node. Makes more sense to merge IFS with global foundries and create a true competitor to TSMC.
 
I still fail to see the logic behind this or why any of the governments who'd need to sign off on it would.

TSMC already has around 67% of all foundry revenue in Q4 (Samsung is second at ~8%) according to TrendForce and that list doesn't include Intel. According to Intel's Q4 2024 financials foundry revenue was $4.5b which is higher than everyone other than TSMC and makes up around 17% of TSMC's revenue for Q4. Just adding Intel in as is (no big contract work yet) would push TSMC a bit over 70% while dropping Samsung to ~7.5%.

Then there's the practical side of this venture requiring guaranteed wafer buys from Intel which as we all know worked out real well for AMD with GloFo.

All this being said given the sheer incompetence of the Intel board and its idiotic makeup I wouldn't put it past them to try.
How does Intel Foundry being added to the pie of Semi Con Foundries cause TSMC Foundry revenue to climb from 67% to over 70%? Are you assuming the JV would be cumulative to their current contributions? I am confused.
 
If Intel sells off and it's a TSMC run venture why wouldn't one count it towards TSMC?
In a joint venture (JV), costs and revenues are shared. The article didn't mention selling Intel Foundry, but rather suggested co-utilization. Do you think Intel could offload the Foundry, especially to a foreign company? I doubt it, but a JV could be possible, although I think it is not strategical beneficial to either company(TSMC helping a competitor, Intel R&D vs TSMC clash). Regardless I don't see how TSMC would claim more than half of the Foundry's Revenue from a new JV. At most, they might realize up to 50% if the JV fully consumes what is currently Intel Foundry.
 
In a joint venture (JV), costs and revenues are shared. The article didn't mention selling Intel Foundry, but rather suggested co-utilization. Do you think Intel could offload the Foundry, especially to a foreign company? I doubt it, but a JV could be possible, although I think it is not strategical beneficial to either company(TSMC helping a competitor, Intel R&D vs TSMC clash). Regardless I don't see how TSMC would claim more than half of the Foundry's Revenue from a new JV. At most, they might realize up to 50% if the JV fully consumes what is currently Intel Foundry.
I think you missed my point entirely:

The article literally says it'd be operated by TSMC which means it would no longer be run by Intel. This also means that it could never be a true competitor to TSMC. So while TSMC wouldn't specifically be taking home all of the revenue they'd still be controlling that amount of the market.
 
  • Like
Reactions: KyaraM and bluvg
I think you missed my point entirely:

The article literally says it'd be operated by TSMC which means it would no longer be run by Intel. This also means that it could never be a true competitor to TSMC. So while TSMC wouldn't specifically be taking home all of the revenue they'd still be controlling that amount of the market.
and then TSMC, the world's largest contract chipmaker, buys less than 50% of IF
TSMC couldn't operate it on its own with only 49% ,51% would still stay with intel.
They would name TSMC as the operator for press reasons but decisions would have to be agreed upon by intel.

Still would only affect the FABs on US soil while intel has FABs in other countries as well.

Also most if not all of intels new FABs are co funded already with intel only owning 51% of them anyway.
https://bip.brookfield.com/press-re...frastructure-signs-definitive-agreement-intel

Also it would take a few years of intel doing as badly as 2024 for intel to even start considering selling off their stake of the FABs.
 
TSMC couldn't operate it on its own with only 49% ,51% would still stay with intel.
They would name TSMC as the operator for press reasons but decisions would have to be agreed upon by intel.

Still would only affect the FABs on US soil while intel has FABs in other countries as well.

Also most if not all of intels new FABs are co funded already with intel only owning 51% of them anyway.
https://bip.brookfield.com/press-re...frastructure-signs-definitive-agreement-intel

Also it would take a few years of intel doing as badly as 2024 for intel to even start considering selling off their stake of the FABs.
Clearly you didn't read either article if you think that's what this is about so here's a quote:
Under the proposal, the Taiwanese chipmaking giant would run the operations of Intel's foundry division
Also if you think Intel retains majority what would other stakeholders have?
 
Clearly you didn't read either article if you think that's what this is about so here's a quote:
Yes, my quote was also a quote...
You can't run a business with 49% , you will be high up on the decision making but 49% of a decision is not final.
Also if you think Intel retains majority what would other stakeholders have?
No matter, even if it's not intel but the other stakeholders, if tsmc has only 49% they can't make any decisions by themselves.
 
"
  • Transferring TSMC's production nodes to Intel's advanced fabs with EUV tools is hard, if possible at all, as Intel and TSMC operate vastly different manufacturing processes, employing distinct equipment configurations and chemical substances at their respective fabs.
"

Intel is now run by Cadence's former CEO who has a long track record of making design tools for TSMC. I'd imagine his #1 priority is to align Intel fabs with industry standard process... and by "industry" standard I really mean TSMC standard.
 
"
  • Transferring TSMC's production nodes to Intel's advanced fabs with EUV tools is hard, if possible at all, as Intel and TSMC operate vastly different manufacturing processes, employing distinct equipment configurations and chemical substances at their respective fabs.
"

Intel is now run by Cadence's former CEO who has a long track record of making design tools for TSMC. I'd imagine his #1 priority is to align Intel fabs with industry standard process... and by "industry" standard I really mean TSMC standard.
That's not the point, intel fabs are made to produce 18a and you can't do much to turn them into fabs that make n3 or whatever tsmc newest node is, other than basically scrap everything except for the buildings and retool them.
But if you add that cost on top of the buying cost...
 
Do I have to to know that 18a and n3 isn't the same thing?!
No, I was just wondering given the context. Seems to me like you have insider information because I cannot verify anything you say elsewhere. We have argued about stuff regarding Intel before and I have read many of your arguments in the forums. I can only explain your arguments at this point by assuming you know stuff people outside Intel do not.
 
Last edited:
Sure, if 'run the business' means that every decision they make can be vetoed by the rest of the owners....
You don't have a clue how publicly traded businesses work if you think this is a barrier to operating a company. Even if their goal is for it being a private venture it still wouldn't slow them down as it would take everyone else opposing something they're doing to stop them.

Most likely the "49%" thing from TSMC is a ploy to get potential regulators to approve the deal. This hardly precludes them from having a small stakeholder who proxies for them.
 
Last edited:
  • Like
Reactions: helper800
There are many comments about whether it's in TSMC's financial interest to be involved in a JV with a competitor. However, there might be important geopolitical factors involved in addition to the financial issues. Considering the ever more ominous threat to Taiwan from the CCP, a threat which the USA is no longer able to counter, such factors might be significant ones in TSMC's decision making.
 
That's not the point, intel fabs are made to produce 18a and you can't do much to turn them into fabs that make n3 or whatever tsmc newest node is, other than basically scrap everything except for the buildings and retool them.
But if you add that cost on top of the buying cost...
Neither is TSMC N3 directly compatible with N2, but there is a path to port designs. Im not saying TSMC would come it and replace all Intel's equipment and processes with their own, but i am saying Intel needs a *design* process that is portable with TSMC's. With the hiring of Cadence's former CEO, that's now a forgone conclusion in my mind.