Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/has-intel-stumbled-again.15965/
        )

    [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
)

Has Intel Stumbled Again?

fansink

Active member
Intel is evaluating a revision to its blueprints for the 14th Gen Core "Meteor Lake" CPUs by turning to TSMC's 5nm process family to manufacture all of the integrated chips, according to industry sources.


 
Intel is evaluating a revision to its blueprints for the 14th Gen Core "Meteor Lake" CPUs by turning to TSMC's 5nm process family to manufacture all of the integrated chips, according to industry sources.


Meteor Lake is designed with the Intel Tile (Chiplet) architecture using Intel 4(7 nm). It's not just one single traditional CPU. I don't think Intel can change the planning, design, and manufacturing for Meteor Lake from Intel in-house to TSMC in just one year. Unless it was the plan all along.
 
Last edited:
  • Like
Reactions: VCT
Meteor Lake is designed with the Intel Tile (Chiplet) architecture using Intel 4(7 nm). It's not just one single traditional CPU. I don't think Intel can change the planning, design, and manufacturing for Meteor Lake from Intel in-house to TSMC in just one year. Unless it was the plan all these days.

Since Pat is used to hustling... and since Pat has known what little progress Intel is actually making, Pat has undoubtedly prepared accordingly, else possibly lose his lucrative $180M employment package?
 
Two concerns with this claim --
1. This rumor directly contradicts what Intel said themselves just a few days ago: https://www.anandtech.com/show/17366/intel-meteor-lake-client-soc-up-and-running - "based on Intel 4"
2. There are previous claims the iGPU would use the TSMC N3 process. Perhaps that is what is delayed? .. or perhaps this latest rumor is made up.
N3 for iGPU source (example): https://ctee.com.tw/news/tech/552277.html

1.) “Up and running” and Pat’s famous “manufacturing ready” in no way implies Intel 4 wafers have been fabbed at sufficient volumes to have confidence in yields, especially since current Intel 4 wafers are lab prototype quantities only, due to lack of EUV tools.

2.) Most likely, TSMC’s 4nm and 3nm are booked solid, and 5nm is the only node available to Intel at such short notice (likely negotiated during his last trip).
 
1.) “Up and running” and Pat’s famous “manufacturing ready” in no way implies Intel 4 wafers have been fabbed at sufficient volumes to have confidence in yields, especially since current Intel 4 wafers are lab prototype quantities only, due to lack of EUV tools.

2.) Most likely, TSMC’s 4nm and 3nm are booked solid, and 5nm is the only node available to Intel at such short notice (likely negotiated during his last trip).

Intel and TSMC been negotiating for a very long time. The go in betweens TSMC, and Intel date to Krzanich time, and basically they been all about volume, and cost.

Intel can't believe TSMC does not jump on the opportunity to give them a deal basically.
 
Intel and TSMC been negotiating for a very long time. The go in betweens TSMC, and Intel date to Krzanich time, and basically they been all about volume, and cost.

Intel can't believe TSMC does not jump on the opportunity to give them a deal basically.
In terms of leading edge capacity and capabilities, TSMC has multiple customers who are willing to buy TSMC's service and capable to put down large prepayments to secure their slots. Several of them are bigger than Intel in market cap or revenue.

On the other hand, without working with TSMC I don't see where Intel can find another capable foundry partner that fits Intel's needs.
 
Pat Gelsinger’s grandiose schedule to surpass TSMC by 2025 has obviously slipped closer to reality, and a case of lipstick won’t change that fact.

Exponential growth claims are increasingly difficult to meet, the further flat growth drops below that curve.
 
sad thing is, I trust this "rumor" more than Intel's public statements. Intel just doesn't have much credibility when it comes to things they say.

they claim their next gen nodes are "ahead of schedule" all while booking up all of TSMC's 3nm leftover capacity. sounds legit to me!
 
Architecture and node size are becoming too complex. Mems or an invasion of Taiwan would change their fortune.
 
sad thing is, I trust this "rumor" more than Intel's public statements. Intel just doesn't have much credibility when it comes to things they say.

they claim their next gen nodes are "ahead of schedule" all while booking up all of TSMC's 3nm leftover capacity. sounds legit to me!

There's a lot more to the story than that with TSMC's N3 orders. I have no inside knowledge, and Intel may not be painting an accurate picture on their near term roadmap (I don't honestly know).. but consider a few things:

1. Intel is entering the GPU market again and plans to aggressively ramp/sell ARC over the next few generations. TSMC N3 was planned to be the node to sell millions of these devices in the future (2nd or 3rd generation - Battlemage or Celestial). The quickest way to reduce risk on launching a whole new product type is to outsource production to a reliable partner.

2. Intel has a lot of previous commitments for TSMC node capacity from the previous CEO. The current CEO might still be buying additional TSMC capacity but he's also the one trying to re-jumpstart Intel's engineering efforts - so there's a chance he's literally going to do both - buy a lot of TSMC capacity at advanced nodes, and push Intel forward on manufacturing again.

I also wouldn't count Intel permanently out or behind - they have a really strong history on delivering "first" silicon technologies. There were quite a few long stretches that the foundries have been on the back foot vs. Intel's manufacturing capabilities over the last 30-40 years..
 
People that accomplished something are being poached from others, google, apple, nvidia, ... and it's causing problems.
 
I also wouldn't count Intel permanently out or behind - they have a really strong history on delivering "first" silicon technologies.

Intel has had many technological “firsts”, but there is no doubt they have significantly slipped in the last decade. Can they be resuscitated, yes, but their faded shine will likely continue fading for quite some time.
 
There's a lot more to the story than that with TSMC's N3 orders. I have no inside knowledge, and Intel may not be painting an accurate picture on their near term roadmap (I don't honestly know).. but consider a few things:

1. Intel is entering the GPU market again and plans to aggressively ramp/sell ARC over the next few generations. TSMC N3 was planned to be the node to sell millions of these devices in the future (2nd or 3rd generation - Battlemage or Celestial). The quickest way to reduce risk on launching a whole new product type is to outsource production to a reliable partner.

2. Intel has a lot of previous commitments for TSMC node capacity from the previous CEO. The current CEO might still be buying additional TSMC capacity but he's also the one trying to re-jumpstart Intel's engineering efforts - so there's a chance he's literally going to do both - buy a lot of TSMC capacity at advanced nodes, and push Intel forward on manufacturing again.

I also wouldn't count Intel permanently out or behind - they have a really strong history on delivering "first" silicon technologies. There were quite a few long stretches that the foundries have been on the back foot vs. Intel's manufacturing capabilities over the last 30-40 years..

And there's now been a long stretch where Intel has been on the back foot vs. TSMC manufacturing capacity for more than 5 years...

TSMC overtook Intel by spending more money and pushing bigger volumes of many different designs through the fabs to get yield up and cost down. It's difficult to see how Intel can do the same to take back the process lead, especially with their lack of EUV equipment and experience using it in mass production.
 
It's difficult to see how Intel can do the same to take back the process lead, especially with their lack of EUV equipment and experience using it in mass production.

Since 2019, on average, TSMC has increasingly acquired more EUV tools than Intel, at a rate of 22 EUV tools per year. If this rate continues, TSMC will have 88 more EUV tools by the end of 2022, and 154 more EUV tools than Intel by 2025.
 
Since 2019, on average, TSMC has increasingly acquired more EUV tools than Intel, at a rate of 22 EUV tools per year. If this rate continues, TSMC will have 88 more EUV tools by the end of 2022, and 154 more EUV tools than Intel by 2025.

Agreed, limited EUV throughput, which is why Intel is using TSMC at an increasing rate. In 2025 more than half of the chips Intel ships will be manufactured by TSMC, absolutely.
 
Intel shareholders voted last week against the company's compensation for its top executives. However the vote is only advisory and won't immediately affect the compensation of Intel's executives.

Meanwhile, Intel is handing tech chief Ann Kelleher a 'blank check' in hopes she can lead Intel back to being the leader in technology
 
Intel shareholders voted last week against the company's compensation for its top executives. However the vote is only advisory and won't immediately affect the compensation of Intel's executives.
Meanwhile, Intel is handing tech chief Ann Kelleher a 'blank check' in hopes she can lead Intel back to being the leader in technology

Wow:

Gelsinger took over as CEO of Intel in Feb. 2021, and received a compensation package worth $178.59 million later that year, according to a financial filing. The compensation included over $1 million in salary, a $1.75 million bonus, over $140 million in stock awards, and nearly $30 million in option awards.

Q1 2022 vs. Q1 2021 Revenue
Intel: -7% (-1% non GAAP, -7% GAAP)
 
Intel is handing tech chief Ann Kelleher a 'blank check' in hopes she can lead Intel back to being the leader in technology

Does Ann Kelleher have the knowledge and technological resources to orchestrate the turnaround Pat Gelsinger has promised?
 
Does Ann Kelleher have the knowledge and technological resources to orchestrate the turnaround Pat Gelsinger has promised?
She has the best technical credentials for the job of any Intel exec in recent memory:


As for the technological resources, Daniel and others have already pointed out that Intel will be woefully behind TSMC in EUV capabilities for as far as one can project in the future.
 
Back
Top