Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/intels-problem-is-tsmc-and-it-wont-go-away.20793/
        )

    [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] => 2021370
            [XFI] => 1050270
        )

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

Intel's problem is TSMC and it won't go away

A

Arthur Hanson

Well-known member
Intel's biggest problem is TSMC and it is not only not going away but getting stronger while building new fabs on Intel's home turf. Intel will need some radical changes to get out of a mess caused by arrogance. It could be done, but the right person who can put together a good team will be needed. If anyone has an idea who the right person is, please post the name and the what they would bring to the table.
 
i will just wait one year to see how it turns after that even if they fire him and hire someone it will only cause a short stock uptick might please shareholders the revenue will not change in short term either
 
Besides Intel's former CEOs lacking necessary technology background to make right decisions, I would think Intel's patterning & BEOL teams should take the major responsibility for losing the technology leadership. On the patterning side, did Intel invent or pay for the SAQP technology (i.e., dual-matterial SAQP for BEOL to solve EPE issue, US 9679771 B1) used in their N10 and else products? It may be just a penalty from universal rules.

Dr. Burn Lin, a well-known lithography expert, played a key role in TSMC's catching up. He already retired, but may be available for an enjoyable part-time job for Intel. On the future BEOL, I am afraid that Intel's decision to adopt the BPDN (backside power delivery network) might be too aggressive, just similar to what they did to introduce Co and have to come back to Cu (Co cladding) again to save yield. Company wise, Mark Liu, Lisa Su, Jensen Huang...may be good candidates for Intel.

Nevertheless, I am optimistic about Intel's long-term perspective, it just needs some more time.
 
Last edited:
TSMC was ALWAYS going to take leadership. They are the best company and have the highest volume. It wasn't an Intel mistake. TSMC is allowing intel to be successful at this point.

Intel just needs to execute on becoming cost effective (which they have never been). So far it looks like it is struggling.
 
It is worse than most imagine. TSMC is giving IP companies free shuttles and other perks to make sure the ecosystem is strong. For those who are waiting for TSMC to become a monopoly monsters you are going to be disappointed.
 
Daniel Nenni said:
It is worse than most imagine. TSMC is giving IP companies free shuttles and other perks to make sure the ecosystem is strong. For those who are waiting for TSMC to become a monopoly monsters you are going to be disappointed.
Click to expand...
Interesting. I would guess Intel has been giving away shuttles and wafers and that might be contributing to foundry Margin issue.
TSMC enables all product companies to have leading edge technology with flexible output. This was unheard of 20 years go. People can choose Samsung, UMC, GF, Intel, SMIC anytime they want....
 
MKWVentures said:
Interesting. I would guess Intel has been giving away shuttles and wafers and that might be contributing to foundry Margin issue.
TSMC enables all product companies to have leading edge technology with flexible output. This was unheard of 20 years go. People can choose Samsung, UMC, GF, Intel, SMIC anytime they want....
Click to expand...

Actually, Intel, Samsung, and Rapidus are paying for IP to be ported and/or developed and it is quite the boost for the IP industry (Synopsys, Arm, Cadence, etc...). A bit of a bubble. The same thing happened when Globalfoundries first came to town. Unfortunately, unless Intel and Samsung get customers for that IP it will be short lived. It can cost millions of dollars to port something over but for IP vendors that is NRE, not high margin licensing and royalties. TSMC, on the other hand, is the golden goose when it comes to customers wanting to license commercial IP.
 
Intel has quite a lot of IP gor Internal use as well can't they use that as a + point in their business I know it will be not that much compared to TSMC
 
siliconbruh999 said:
Intel has quite a lot of IP gor Internal use as well can't they use that as a + point in their business I know it will be not that much compared to TSMC
Click to expand...

That was the hope when Intel decided to get back into the foundry business but as it turns out customers want commercial IP that they trust and can use on other foundries. Intel IP is also specific to Intel so it may not be competitive with commercial offerings. If the Intel IP is amazing and unlike others then Intel will keep it internal.
 
My opinion: intel is a poor rich family which means they used to be very rich and now getting poor and want to earn back the glory. Historically, if the family wants to keep the same grandeur and want to win back the glory in a short time, it needs super great luck and focus on core/unique business at beginning. Currently the super great luck for intel could be "China invading Taiwan and all supply chains get disrupted". It seems will not happen in the near future.
It is a pity that intel did not leverage tsmc's 3m/2nm capacity effectively in the past and now. If tsmc takes more than 95% of market share in 5nm/3nm foundry, then 30% capacity could mean >30% market share in the end market for advanced node application in CPU/GPU and others besides mobile chips (Mobile AP might take large percentage in the 5nm/3nm wafer output). If intel creates extra effective capacity in 3nm/2nm, which means he control all the upside capacity...........
It is tough time for intel and it takes time to reshuffle, re-organize and focus the core team.
 
Daniel Nenni said:
That was the hope when Intel decided to get back into the foundry business but as it turns out customers want commercial IP that they trust and can use on other foundries. Intel IP is also specific to Intel so it may not be competitive with commercial offerings. If the Intel IP is amazing and unlike others then Intel will keep it internal.
Click to expand...
Intel have plenty of reason and opportunity to be testing their IP on TSMC right now, since they are making multiple different advanced and critical products there. It makes no sense that they would not include what they believe to be the best IP, and I would expect TSMC to be trusted to keep it confidential as they have long practiced.

So Intel should be in a position to offer portable IP for many interesting things, satisfying their customers that they are not locked in.
 
Daniel Nenni said:
Actually, Intel, Samsung, and Rapidus are paying for IP to be ported and/or developed and it is quite the boost for the IP industry (Synopsys, Arm, Cadence, etc...). A bit of a bubble. The same thing happened when Globalfoundries first came to town. Unfortunately, unless Intel and Samsung get customers for that IP it will be short lived. It can cost millions of dollars to port something over but for IP vendors that is NRE, not high margin licensing and royalties. TSMC, on the other hand, is the golden goose when it comes to customers wanting to license commercial IP.
Click to expand...
Thanks. Good info.
 
Arthur Hanson said:
Intel's biggest problem is TSMC and it is not only not going away but getting stronger while building new fabs on Intel's home turf. Intel will need some radical changes to get out of a mess caused by arrogance. It could be done, but the right person who can put together a good team will be needed. If anyone has an idea who the right person is, please post the name and the what they would bring to the table.
Click to expand...
It needs not just only right person, but need a lot of right personssssssss, I.e right culture. Without a right culture, it cannot sustain an empire, a company, a family for more than 3 decades; even tsmc. Historically it is always true, we see IBM, GE, Sony, Nokia etc, who can stay at high peak for 3 decades? Either gradually die, or painfully reborn. The only enemy of tsmc is itself, the poisonous culture within itself like the other companies. So we will see the result 10 years later.
 
Arthur Hanson said:
Intel's biggest problem is TSMC and it is not only not going away but getting stronger while building new fabs on Intel's home turf. Intel will need some radical changes to get out of a mess caused by arrogance. It could be done, but the right person who can put together a good team will be needed. If anyone has an idea who the right person is, please post the name and the what they would bring to the table.
Click to expand...

If there are 10 reasons for Intel to compete against TSMC, there will be 100 reasons for Intel to collaborate with TSMC in order to win. The Intel Foundry venture is a death trap designed by Intel itself and cheered by Intel's true competitors.

Intel can do several things to turn its business around but Intel shouldn't treat TSMC as a competitor or even try to compete against TSMC in the foundry market. To compete against TSMC in the foundry market is a wrong move that placed those precious Intel resources and management attention onto a wrong target.

If I am a strategy planning person working for Nvidia, AMD, Broadcom, Qualcomm, MediaTek, or even Apple, I'd suggest that the management should keep giving Intel Foundry positive and encouraging comments. Because it can distract Intel from competing against my company's products and stretch Intel so thin that Intel can't compete against my company.

Also my company can use Intel Foundry as a barging chip to negotiate a better contract term with TSMC. Why not? It's Intel spending Intel's own money to bring down Intel itself!

Unfortunately (or fortunately if you are Intel's competitors), we are watching Intel spent billions of dollars to walk willingly and proudly into this death trap.
 
Besides other setbacks by Intel, the biggest might be its total failure of its GPU project, both on the consumer side and especially on the data center side, zero market share, zero big sales, the Intel GPU team is literally running by monkeys.
 
You must log in or register to reply here.
Back
Top