Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/the-semiconductor-survival-battle-part-1-of-5-lip-bu-tan-intel-vs-c-c-wei-tsmc.23919/page-2
        )

    [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] => 2030770
            [XFI] => 1060170
        )

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

THE SEMICONDUCTOR SURVIVAL BATTLE Part 1 of 5 Lip-Bu Tan (Intel) vs. C.C. Wei (TSMC)

It was more like the US government demanded advanced chip production in the US as a milestone, but Intel already had already previously decided to ramp Intel 4/3 in Ireland first. The team and tools were there. This disqualified Intel from the government subsidy.

Since the Irish ramp was before the EU started giving funding for chips, they got no money from there either.

The US government should have considered this. Unlike TSMC, Intel does process development in the US. In fact it is the only company doing so. I am purposely ignoring IBM here since they produce nothing with it.

Pat cancelled the German, Israeli, and Polish fabs before any construction started. So you can't claim much money was lost there. If there is any millstone it is the Ohio fab.
 
Last edited:
There was no stock buy back under Gelsinger...
And he did build fabs just was not able to fill them... Cause surprise your predecessor had already planned everything at TSMC and you can't shift it back instantly.
 
Last edited:
dkr1986 said:
That very well could be the reason. But we have already started seeing perf leaks like we saw prior to LNL EEP launch between TechTour (June 2024) & product launch (September 2024). Also we are seeing Panther lake based laptops being showcased for public.

https://twitter.com/x/status/1983844466114097652

So I think If they wanted, they could have done limited volume launch for this holiday season with EEP (probably not very good margin wise as volume is low). But head count reductions must have eaten into the device driver readiness and other supporting stuff for a good launch etc. Again just my speculation.
Click to expand...
so that looks like a very expensive processor.
Product is not mature enough for multiple sku launch. one sku to start is a great idea. I believe this was the plan since beginning of the year.
 
20A/18A loadings are 1/4 the amount Pat planned for in 2021 after he shifted the roadmap. That is the reason for fab cancellation. No one at Intel other than Pat thought they would fill the fabs. the balance sheet got too out of hand and it needed to be stopped. Intel built one of 6+ fabs planned and it is ramping slowly, not at max ramp speed. lets see how that goes.
 
MKWVentures said:
20A/18A loadings are 1/4 the amount Pat planned for in 2021 after he shifted the roadmap. That is the reason for fab cancellation. No one at Intel other than Pat thought they would fill the fabs. the balance sheet got too out of hand and it needed to be stopped. Intel built one of 6+ fabs planned and it is ramping slowly, not at max ramp speed. lets see how that goes.
Click to expand...
Do you think if they have focused only on 2 Fabs instead of 6+ they would have been in better position?
 
siliconbruh999 said:
Do you think if they have focused only on 2 Fabs instead of 6+ they would have been in better position?
Click to expand...

Yes, it would reduce a lot of financial stress and trouble if Intel had focused on only two fabs. But that wouldn’t solve Intel’s fundamental product problems. Intel doesn’t have competitive products (or in some cases, any products at all) in those high-growth, high-volume, and high-profit markets.

Intel’s product division must find ways to grow its business with or without a competitive Intel Foundry.

Pat Gelsinger knew he needed to fix Intel’s product problems, but he failed. Lip-Bu Tan knows it too. Without a healthier Intel product division generating huge revenue and profit (regardless of whether the chips are manufactured internally or externally) Intel Foundry won’t be able to compete either.
 
hist78 said:
Yes, it would reduce a lot of financial stress and trouble if Intel had focused on only two fabs. But that wouldn’t solve Intel’s fundamental product problems. Intel doesn’t have competitive products (or in some cases, any products at all) in those high-growth, high-volume, and high-profit markets.
Click to expand...
That is not a IFS issue it's a product issue.
hist78 said:
Pat Gelsinger knew he needed to fix Intel’s product problems, but he failed. Lip-Bu Tan knows it too. Without a healthier Intel product division generating huge revenue and profit (regardless of whether the chips are manufactured internally or externally) Intel Foundry won’t be able to compete either.
Click to expand...
I am sorry but the product that were actually initiated and improved upon pat are yet to launch all Products till Launched till pat was CEO was defined under Swan. ARL was 2019-2020 definition so if you wanna see the impact of Intel products wait for PTL/CLW-F/DMR/NVL judge him based on them.
 
siliconbruh999 said:
Do you think if they have focused only on 2 Fabs instead of 6+ they would have been in better position?
Click to expand...
No. Fabs are a outcome of customers requirements. No recovery in market share, No AI GPUS was part one. Part two was that somehow 50+ people looked at 20A, 18A, Intel 3 and all said "no thanks". In 2021 Intel literally expected 2+ fabs just for external foundry in 2026. Total actual needed is zero.

What they should have done.... and what they are doing now... is customer looks at the technology, make a committment, and we will provide capacity. The total capacity people are even potentially looking at is <1 fab in 2028+ timeframe so Intel can fit it in if needed.

Finances are still a problem ....Only way IFS breaks even in 2027 is if :

Write off 20B+ in assets to reduce depreciation in future years
Charge product group higher wafer prices (IFS looks good, Product looks bad)

Reminder: IFS Operating margin is -50% when charging TSMC price. TSMC margin is 50% when chargin TSMC price. You can calculate Intel cost from this. EUV did not improve the margin.
 
siliconbruh999 said:
That is not a IFS issue it's a product issue.

I am sorry but the product that were actually initiated and improved upon pat are yet to launch all Products till Launched till pat was CEO was defined under Swan. ARL was 2019-2020 definition so if you wanna see the impact of Intel products wait for PTL/CLW-F/DMR/NVL judge him based on them.
Click to expand...

Without Intel’s product divisions selling the right products (whether manufactured internally or externally) to bring in strong revenue and profit, how can Intel afford to invest billions of dollars into Intel Foundry?

Yes, many product development efforts began long before Pat Gelsinger returned to Intel. But he has been CEO for almost four years. If Intel’s product shortcomings today have nothing to do with his strategy and execution over that four years, then why does Intel need him? And looking ahead, could Lip-Bu Tan make the same excuse to blame Pat Gelsinger in 2029 if Intel’s product business still underperforms?

A CEO ultimately bears responsibility and takes credit for a company’s success or failure. It has been nearly five years since Pat Gelsinger became CEO, almost one year since he stepped down, and about eight months since Lip-Bu Tan became the new CEO. Yet Intel still lacks competitive products in the red hot AI market. In my view, Pat Gelsinger did not guide Intel effectively into this AI revolution, or he tried, but failed.
 
sit said:
Disagree. The fab construction binge had its roots in a hyper-optimistic misunderstanding of how much and how fast foundry business would come to Intel plus the perverse incentive of getting gov’t money for building something not needed.

The of the 4 fab projects (AZ, Ohio, Israel, and Germany) only one was needed though 3 continue albeit at snails paces.

At current growth rates Intel will be sitting on 20 to 30 years of fab space. A huge millstone.
Click to expand...
No, the truth is that the government was originally going to provide subsidies but refused to provide them.
It's not just Intel
 
Honestly, it's a bit puzzling that the government didn't implement the CHIPS Act.
It's not that Intel betrayed the government, but rather that the government betrayed the companies that it promised to subsidize through the CHIPS Act...
 
hist78 said:
Yes, this can be interpreted in two different ways.

My view:
The US government didn’t fully trust that Intel would refrain from using taxpayer funded support for share buybacks. Therefore, the restriction was explicitly written into the final agreement to prevent Intel from doing so.

Your view:
Since both the US government and Intel had already agreed not to use these taxpayers' money for share buybacks, the restriction was included in the final agreement for formality and transparency.



Based on Pat Gelsinger’s public comments in 2023 and 2024, it was clear that Intel was not pleased with the restrictions and milestone based structure of the CHIPS Act funding. Intel’s final funding agreement was also noticeably delayed.

In contrast, we did not hear much from TSMC, GlobalFoundries, Micron, SK Hynix, Texas Instruments, Samsung, Analog Devices, or other semiconductor companies complaining about the CHIPS Act requirements. In fact, shortly after the 2024 US presidential election, the Biden Administration announced the final funding agreement with TSMC on November 15, 2024. Eleven days later, on November 26, 2024, Intel’s final funding agreement was announced.

Given the timing, one could reasonably assume that there were significant disagreements on several issues between Intel and the US Department of Commerce. It appears Commerce Department could no longer delay announcing the long awaited agreement with TSMC, even though TSMC is a foreign company.
Click to expand...
No, it's just that the government was slow to respond.
 
tomatoma said:
Honestly, it's a bit puzzling that the government didn't implement the CHIPS Act.
It's not that Intel betrayed the government, but rather that the government betrayed the companies that it promised to subsidize through the CHIPS Act...
Click to expand...

Not all companies, just the ones who could not do what they signed up to do. TSMC got their money, yes?
 
siliconbruh999 said:
It's mostly Intel 7 domination RN we will have to see what happens when they moved volume to 18A
Click to expand...
the margins are better on 7 than 3 or 4. and we already heard about 18A being a headwind in 2026 from DZ.

Volume leads to improved margins. Cant be successful spending tons on money on a technology that barely fills one HVM fab sometime in 2027
 
You must log in or register to reply here.
Back
Top