Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/tsmc-faces-a-capacity-shortage-leading-six-major-foundries-to-choose-samsung-over-intel-for-manufacturing.24383/
        )

    [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
)

TSMC faces a capacity shortage, leading six major foundries to choose Samsung over Intel for manufacturing?

jiakun0618

New member
TSMC faces a capacity shortage, leading six major foundries to choose Samsung over Intel for manufacturing?
With AI driving up chip demand and TSMC 's capacity remaining fully utilized, Deutsche Bank observes that AMD, Nvidia, Qualcomm, and Apple are now urgently seeking second suppliers. Samsung Electronics Co. has become the preferred outsourcing partner for these tech giants, as it is more attractive than Intel Corp.'s wafer foundry services.
On the 19th, well-known tech leaker Jukan pointed out through the social media platform X that a research report recently released by a team of analysts led by Robert Sanders at Deutsche Bank indicated that TSMC's 3nm production capacity is currently extremely tight. Not only is it sold out in 2026, but orders are even backlogged into 2027, forcing TSMC to significantly increase its capital expenditure in 2026 to US$52-56 billion, higher than Deutsche Bank's original forecast of US$50 billion and the market consensus of US$46 billion.
This shows that TSMC's capital expenditure plan for 2025 is "too conservative" compared to the explosive demand for AI. The current situation is not just a shortage of CoWoS packaging capacity, but also a serious shortage of core wafer manufacturing (especially 3nm process).
Despite the mixed track record of Samsung and Intel's foundry services, Deutsche Bank points out that Apple, Nvidia, AMD, Broadcom, Qualcomm, and MediaTek now have no other choice. This means that TSMC's market share in the advanced process foundry market is expected to drop from 95% to 90%.

According to Deutsche Bank analysis, Samsung's Taylor, Texas foundry (SF2P) may be more attractive than Intel's. "For customers looking for alternative supply sources, Samsung's Taylor plant is probably their first choice."
The report states that Qualcomm and AMD are the most likely customers to consider Samsung, while Apple and Broadcom are reportedly evaluating Intel's technology. However, Deutsche Bank believes that while Intel's 14A process does have potential, "there is still much work to be done."
https://technews.tw/2026/01/20/tsmc-capacity-rumor/
 
TSMC faces a capacity shortage, leading six major foundries to choose Samsung over Intel for manufacturing?
I am doing a report on these shortages.... trying to get the numbers to make sense... please chime in to correct me.

1) If you are choosing a wafer foundy for a design, it is 2-3 years minimum from this decision to actually needing volume. So if I have a design, and I decide to do it at Samsung vs TSMC, it means TSMC doesnt have capacity in 2-3 years.

2) If I am choosing a wafer foundry based on on FEAR of shortage or priority, or because TSMC turned me away, this makes sense. I would be careful about choosing Intel or Samsung because you think they will deliver better in 2-3 years.

3) given the demand for TSMC, they will choose the best customers to support and not support others .... which may be leading them to go to other companies.

from what I have heard EVERYONE is evaluating Intel's technology. Just like they did in 2021-2023. The question is who has committed to it and whether it will ramp.

Question: What is Intels competitive advantage over Samsung in foundry. What makes Intel or Samsung a better choice?
 
I am doing a report on these shortages.... trying to get the numbers to make sense... please chime in to correct me.

1) If you are choosing a wafer foundy for a design, it is 2-3 years minimum from this decision to actually needing volume. So if I have a design, and I decide to do it at Samsung vs TSMC, it means TSMC doesnt have capacity in 2-3 years.

2) If I am choosing a wafer foundry based on on FEAR of shortage or priority, or because TSMC turned me away, this makes sense. I would be careful about choosing Intel or Samsung because you think they will deliver better in 2-3 years.

3) given the demand for TSMC, they will choose the best customers to support and not support others .... which may be leading them to go to other companies.

from what I have heard EVERYONE is evaluating Intel's technology. Just like they did in 2021-2023. The question is who has committed to it and whether it will ramp.

Question: What is Intels competitive advantage over Samsung in foundry. What makes Intel or Samsung a better choice?

One big advantage: Intel 18AP and 14A will be made in America by an American company that the US Government is a major stakeholder.

TSMC's $50B+ CAPEX should tell you something. It takes 2 years to build a fab in Taiwan. Same with a packaging facility. Since it takes 2-3 years to tape-out a chip TSMC is right on schedule.

The problem with capacity is that it is built based on signed wafer agreements. For N3 customers are buying more than they originally signed up for. TSMC is moving N5 fabs to N3 as a result. Sometimes it is the other way around but with AI capacity will be tight for sure. TSMC is moving N5 fabs to N3 as a result, they are interchangeable.

Now that AI is more predictable I do not see a capacity problem for TSMC N2. TSMC know it has 99% of the design wins so they will build accordingly.
 
One big advantage: Intel 18AP and 14A will be made in America by an American company that the US Government is a major stakeholder.
I agree on the other timeline comments 100%... that my point

I dont think that having american company build in america with government ownership is necessarily a great deal unless everything else is equal. I think that the best product at the best price from the best company is more important. But I am a free trader and I don't fit into the "new conventional wisdom". Lets see how this new focus ages over the next two years.
 
Back
Top