Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/intel-reports-second-quarter-2025-financial-results.23234/page-7
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Intel Reports Second-Quarter 2025 Financial Results

Daniel Nenni said:
Customers must fund this, not investors. If customers want a second source here in the United States Intel Foundry will succeed. If customers are comfortable with a single source out of Taiwan then so be it. I believe 75% of TSMC's business is from North America so it is in everyone's best interest for Intel Foundry to succeed.
Click to expand...

Still do not understand the "obsession" of some people with "second foundry sourcing" in USA. INTEL for some time during the 2010-2020 thought they needed to second source Lithography tools, so they kept Nikon "kind of alive", but many people saw that this was a lost cause. TSMC never bothered the second sourcing obsession and knew that tight integration with their dominant Lithography supplier was giving them much more added value in running their fabs, flexibly planning increasing/adapting capacity (expansion) when they needed that etc. Currently, TSMC is rapidly expanding/ramping foundry capacity in USA, so that in 4-5 years 30% of their leading edge production capacity, is in USA. Isn't that "second source capacity" enough for you in USA? The whole ecosystem of foundry suppliers is being extended/copied to Arizona, if the local (US) work force is now also learning/adapting to how to run a foundry in the 21st century, the 20% extra costs the customers of TSMC pay are simply worth the value added by TSMC.

Just accept that the world is always changing and that INTEL is not ruling the semi-waves any more in the world, certainly not in manufacturing/foundry. Economy of scale seems to be the determining issue here in combination with TRUST, TRUST and TRUST, customers simply like the flexibility, added value in state-of-the-art packaging and the trust offered by TSMC.

When TSMC's gross margin stays somewhere between 53-60% customers should be happy with the enormous amount of capital TSMC is investing every year in extending leading edge capacity in USA and Taiwan, besides the specialty (automotive, sensor) foundry capacity in Europe and Japan.

Just my two cents....
 
user nl said:
Still do not understand the "obsession" of some people with "second foundry sourcing" in USA. INTEL for some time during the 2010-2020 thought they needed to second source Lithography tools, so they kept Nikon "kind of alive", but many people saw that this was a lost cause. TSMC never bothered the second sourcing obsession and knew that tight integration with their dominant Lithography supplier was giving them much more added value in running their fabs, flexibly planning increasing/adapting capacity (expansion) when they needed that etc. Currently, TSMC is rapidly expanding/ramping foundry capacity in USA, so that in 4-5 years 30% of their leading edge production capacity, is in USA. Isn't that "second source capacity" enough for you in USA? The whole ecosystem of foundry suppliers is being extended/copied to Arizona, if the local (US) work force is now also learning/adapting to how to run a foundry in the 21st century, the 20% extra costs the customers of TSMC pay are simply worth the value added by TSMC.

Just accept that the world is always changing and that INTEL is not ruling the semi-waves any more in the world, certainly not in manufacturing/foundry. Economy of scale seems to be the determining issue here in combination with TRUST, TRUST and TRUST, customers simply like the flexibility, added value in state-of-the-art packaging and the trust offered by TSMC.

When TSMC's gross margin stays somewhere between 53-60% customers should be happy with the enormous amount of capital TSMC is investing every year in extending leading edge capacity in USA and Taiwan, besides the specialty (automotive, sensor) foundry capacity in Europe and Japan.

Just my two cents....
Click to expand...
Where do you get the 30% from?

 
Last edited:
Daniel Nenni said:
Customers must fund this, not investors. If customers want a second source here in the United States Intel Foundry will succeed. If customers are comfortable with a single source out of Taiwan then so be it. I believe 75% of TSMC's business is from North America so it is in everyone's best interest for Intel Foundry to succeed.
Click to expand...
Intel shareholders have already made significant contributions to this with almost no benefit—arguably even negative returns. No more. Customers must pay for it, i.e., 14A.

resized.png
 
user nl said:
TSMC is rapidly expanding/ramping foundry capacity in USA, so that in 4-5 years 30% of their leading edge production capacity, is in USA. Isn't that "second source capacity" enough for you in USA?
Click to expand...
Not in my opinion. TSMC process development is still entirely in Taiwan. Worrying just about fabs is very roughly like worrying about where German car manufacturers' final assembly plants are. Just because BMW's largest assembly plant is in South Carolina and Mercedes has a huge assembly plant in Alabama does not mean South Carolina or Alabama are capable of engineering high-end vehicles, or independently constructing them. Most of the complex sub-assemblies (like engines and transmissions) still come from Germany. SC and AL have perfected great paint shops though. :)

A bit of the preceding comment is tongue-in-cheek, because it takes a lot more engineering process to run a chip fab and keep it running than it does a vehicle assembly plant, but I think the rough analogy is still valid. In the case of TSMC, the real innovation and perfecting the fabrication process is still in Taiwan, just like Intel's is in Oregon.

So the TSMC fabs in the US are not a real second source in the business sense (for leverage against monopoly pricing or single-source execution risk reduction), or in the geo-political sense. If Taiwan gets conquered, or willingly joins China, the US fabs won't be much help for very long, unless the Chinese want them to be.
 
blueone said:
Not in my opinion. TSMC process development is still entirely in Taiwan. Worrying just about fabs is very roughly like worrying about where German car manufacturers' final assembly plants are. Just because BMW's largest assembly plant is in South Carolina and Mercedes has a huge assembly plant in Alabama does not mean South Carolina or Alabama are capable of engineering high-end vehicles, or independently constructing them. Most of the complex sub-assemblies (like engines and transmissions) still come from Germany. SC and AL have perfected great paint shops though. :)
Click to expand...
I agree with this in principle, but consider that fab scaling is slowing down significantly. A new "full node benefit" is 15% Perf per Watt. If you're 1-2 "nodes" behind you're not exactly out of the ballpark these days. Yes your products are less competitive, but you have the nice cost benefit an older and mature node at that stage. The US could always fund a moonshot effort like Rapidus to get back to a consistent N-1 (or slightly better).

Of course keeping the TSMC-built US fabs running without Taiwan is a totally different discussion. Maybe OK short term but sounds nearly impossible long term given the complexity of materials required.
 
Xebec said:
I agree with this in principle, but consider that fab scaling is slowing down significantly. A new "full node benefit" is 15% Perf per Watt. If you're 1-2 "nodes" behind you're not exactly out of the ballpark these days. Yes your products are less competitive, but you have the nice cost benefit an older and mature node at that stage. The US could always fund a moonshot effort like Rapidus to get back to a consistent N-1 (or slightly better).

Of course keeping the TSMC-built US fabs running without Taiwan is a totally different discussion. Maybe OK short term but sounds nearly impossible long term given the complexity of materials required.
Click to expand...
Many of the key consumables comes from outside the US, they may have a local EBO but in reality if you to a tapein it is impossible without Taiwan… so manufacturing old Or existing yes, new products can no do
 
Xebec said:
I agree with this in principle, but consider that fab scaling is slowing down significantly. A new "full node benefit" is 15% Perf per Watt. If you're 1-2 "nodes" behind you're not exactly out of the ballpark these days. Yes your products are less competitive, but you have the nice cost benefit an older and mature node at that stage. The US could always fund a moonshot effort like Rapidus to get back to a consistent N-1 (or slightly better).
Click to expand...
This is a different but similarly interesting discussion. I've always thought that Intel's previous fab scaling leadership made the company too risk-averse to the architecture and design improvements that can be much more of a differentiation than fab scaling. (In fairness, Intel did try that once, with Itanium, but that's such a lousy example, and was such a controversial decision, that I almost hate to mention it.) The one shining example I can think of where an N-2 chip has unchallenged leadership is the Cerebras WSE-3, which is on TSMC 5nm process. The ultimate in thinking different, IMO. Design innovation can be more important than fab process.
Xebec said:
Of course keeping the TSMC-built US fabs running without Taiwan is a totally different discussion. Maybe OK short term but sounds nearly impossible long term given the complexity of materials required.
Click to expand...
My point exactly.
 
Xebec said:
Can Taiwanese manufacturing or Chinese manufacturing exist without the US?
Click to expand...
As much as anyone here, I suspect, you know the answers to this question. I think the answer is complicated. For TSMC as it exists today. it's a hard NO. In the Chinese case the answer has a bunch of expensive caveats that go with it, since they are willing to tolerate low yields, and they are willing to pirate software tools, and they are willing to use older or less capable equipment. And they can do this because they have a state-managed market economy. The world TSMC lives in is not state-managed.
 
In case of China taking over Taiwan, does the US think they can transport/smuggle out enough TSMC brains to run the TSMC fabs here standalone? One can argue that scheme worked with Von Braun and the other German rocket designers, and many European nuclear brains. Is that what US admins (not just the current one) have in their calculus???

It would be better imo if the US really committed to rebuilding chip tech, rather than going on some crash course Manhattan style project when/if the s*** hits the fan in TW.
 
Markwrob said:
In case of China taking over Taiwan, does the US think they can transport/smuggle out enough TSMC brains to run the TSMC fabs here standalone? One can argue that scheme worked with Von Braun and the other German rocket designers, and many European nuclear brains. Is that what US admins (not just the current one) have in their calculus???

It would be better imo if the US really committed to rebuilding chip tech, rather than going on some crash course Manhattan style project when/if the s*** hits the fan in TW.
Click to expand...
don't worry they will realize it too late and are gonna pay the price also they need to support micron as well along with TI gotta have leading/lagging logic memory
 
blueone said:
SC and AL have perfected great paint shops though. :)

A bit of the preceding comment is tongue-in-cheek, because it takes a lot more engineering process to run a chip fab and keep it running than it does a vehicle assembly plant, but I think the rough analogy is still valid. In the case of TSMC, the real innovation and perfecting the fabrication process is still in Taiwan, just like Intel's is in Oregon.
Click to expand...

Hard to understand then why INTEL cannot run your "paint shop"........
 
XYang2023 said:
https://twitter.com/x/status/1948978702349668543
Click to expand...

Agentic AI has strong requirements for CPU and memory capabilities, as real-world agent use cases often involve web operations (e.g., booking an air ticket, chatting with a customer). I believe Intel and AMD could challenge NVIDIA from this angle.

While NVIDIA's superchips already bundle ARM-based CPUs, these may be inadequate in terms of cpu power. I hope that when designing racks, Intel can provide greater flexibility in the ratio between GPUs and CPUs, so that customers could change this ratio in a realtively easy way (AI workloads really change quite a bit from time to time).

Btw, Xyang, could you please relay the above message to X?
 
kevin01 said:
Agentic AI has strong requirements for CPU and memory capabilities, as real-world agent use cases often involve web operations (e.g., booking an air ticket, chatting with a customer). I believe Intel and AMD could challenge NVIDIA from this angle.

While NVIDIA's superchips already bundle ARM-based CPUs, these may be inadequate in terms of cpu power. I hope that when designing racks, Intel can provide greater flexibility in the ratio between GPUs and CPUs, so that customers could change this ratio in a realtively easy way (AI workloads really change quite a bit from time to time).

Btw, Xyang, could you please relay the above message to X?
Click to expand...
I think the most compelling case for Intel and agents is access to databases, code running in containers, and files in the cloud. When they completely replace excel and Bi tools that access those today, they will be needing a lot of cpu power.
 
Markwrob said:
In case of China taking over Taiwan, does the US think they can transport/smuggle out enough TSMC brains to run the TSMC fabs here standalone? One can argue that scheme worked with Von Braun and the other German rocket designers, and many European nuclear brains. Is that what US admins (not just the current one) have in their calculus???

It would be better imo if the US really committed to rebuilding chip tech, rather than going on some crash course Manhattan style project when/if the s*** hits the fan in TW.
Click to expand...
The US, meaning Intel in this context, really can run chip tech. As Dan often points out, Intel has been the source of numerous chip technology advancements in the not distant past. Intel's problem has been a lack of competent leadership for years now, not a lack of engineering capability. I still am not seeing brilliant leadership in Intel at all.

As multiple people have explained here, TSMC fabs cannot operate in the long term without the groups in Taiwan. A few, or even many, transplants will not enable the US to run the TSMC fabs long term without the rest of TSMC in Taiwan. And that's not even taking into account independent suppliers in Taiwan. I think there are two possibilities. One, US government leadership do not understand that fabs need a massive research and development team behind them, and that TSMC is not putting those teams in the US, ever, as far as I can tell. Second possibility is that the US government leadership knows this, and doesn't want to deal with the reality of the situation. I'm going to give them the benefit of the doubt, that they do know just the fabs are not enough, but don't have a story for reality, which would mean energizing Intel. What a mess.
 
blueone said:
The US, meaning Intel in this context, really can run chip tech. As Dan often points out, Intel has been the source of numerous chip technology advancements in the not distant past. Intel's problem has been a lack of competent leadership for years now, not a lack of engineering capability. I still am not seeing brilliant leadership in Intel at all.

As multiple people have explained here, TSMC fabs cannot operate in the long term without the groups in Taiwan. A few, or even many, transplants will not enable the US to run the TSMC fabs long term without the rest of TSMC in Taiwan. And that's not even taking into account independent suppliers in Taiwan. I think there are two possibilities. One, US government leadership do not understand that fabs need a massive research and development team behind them, and that TSMC is not putting those teams in the US, ever, as far as I can tell. Second possibility is that the US government leadership knows this, and doesn't want to deal with the reality of the situation. I'm going to give them the benefit of the doubt, that they do know just the fabs are not enough, but don't have a story for reality, which would mean energizing Intel. What a mess.
Click to expand...

The USA does not control the so-called rare-earth minerals. When China "squeezes DT's balls", like they did for a couple of months, DT caves, he simply has too. The Chinese prepared for this for decade, ask Japan. The USA, it's military, it's automobile industry, it's robotics industry and many other industries, like in the rest of the world, would come to a standstill within 6-12 months. After China squeezed for a couple of months, DT caved to China. China (and others) know how to "play DT".

Read this story of today, DT is so simple-minded (I always stay away from people who think/speak of themselves that they are the smartest and best in the world!!) and most of the people he hired around him are so simple, short term thinking. No trust, independent thinking and strategic/tactical competence in those people that he selects and that are willing to work under him, pure fifth-grade people:

https://www.theguardian.com/us-news...7/trump-deals-trade-economy?CMP=share_btn_url
 
You must log in or register to reply here.
Back
Top