Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/can-intel-regain-process-leadership.12624/
        )

    [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
)

Can Intel Regain Process Leadership?

Daniel Nenni

Admin
Staff member
An article on Seeking Alpha "inspired" by Scott Jones article:

Can TSMC Maintain Their Process Technology Lead

In the article, Jones concludes:

TSMC took the process density lead this year with their 5nm process. Depending on the exact timing of Intel's 7nm process versus TSMC 3nm Intel may briefly regain a process density lead but TSMC will quickly pass them with their 3nm process with over 300 million transistors per millimeter squared!
 
I think it'll be a challenge for Intel. They have struggled to release nodes on schedule and they have struggled even more to ramp them. They have gotten in the habit of claiming a node is ready when the reality they have only managed to manufacture a handful to small chips with terrible yield. Intel's definition of "released" these days is behind what the rest of the industry would define as "risk production". So even if Intel does technically release 7nm in 2021 as planned, I doubt they will be anywhere close to HVM on it for another year or two.
 
Functionality is more important than process leadership. The cloud or mobileye isn't as fashionable but both make billions.

I think we should think of process leadership as part of functionality. When Intel was at 60 or 40 nm the smaller node size meant a more functional product at 14 nm it become more about architecture. Look at mediatek making a killing but they're not part of the process leadership discussion.
 
Last edited by a moderator:
Functionality is more important than process leadership. The cloud or mobileye isn't as fashionable but both make billions.

I think we should think of process leadership as part of functionality. When Intel was at 60 or 40 nm the smaller node size meant a more functional product at 14 nm it become more about architecture. Look at mediatek making a killing but they're not part of the process leadership discussion.

I agree that it's the features or functionality of the product that ultimately determines success and there are many markets, such as embedded, where process leadership isn't extremely important. However Intel's cash cow are the data centric market, which is performance sensitive. Intel does have some other features (and branding) that has helped it hold on to most of it's market share in servers, but are now at a point where TSMC has almost a 2x density advantage. Well performance costs die space. Features cost die space. Intel is going to be forced to make some really difficult tradeoffs to stay competitive.
 
Sales guidance is down 7%. A lot of the products celeron, Pentium .... I wouldn't want. The i5 or i7 I would take a look at but it's expensive and in 6 months may be ordinary. It may be better to wait to get something better for less money. The Lenovo duet is sold out and that's only the start. Word is that hours are being cut at rohler acres.

If it wasn't for the mobileye acquisition working out and the sense I get was a brilliant acquisition and the cloud Intel would be lost. Intel may become a very "boring" company that makes billions in data centers and automation.
 
I believe that Intel needs to start thinking with the mentality that it has not the process lead. Once it starts comprehending this reality, we'll see the company rising again. Now, it seems stuck in that and it is very slow to react. There are situations that having the process lead makes all the sense in the world - for example trying to compete in the high core count arena, becomes difficult. However as companies that try to compete with Intel but until recently thought they would be in the very best case at process parity made interesting decisions (e.g. chiplets, multiple threads per core, accelerators etc etc). Intel should take an example and try to balance the architecture with what you can reasonably expect by the implementation. On the other hand, as previous commenters correctly pointed out, there are a lot of situations that bleeding edge is not as crucial and Intel should make the most of it, since it has quite an arsenal of products that do not really depend on having the most dense process (e.g. chipsets, wireless/wired networking, controllers, power electronics etc) and maybe these products could be a source of future growth more than what high-performance CPUs are now.
 
I believe that Intel needs to start thinking with the mentality that it has not the process lead. Once it starts comprehending this reality, we'll see the company rising again. Now, it seems stuck in that and it is very slow to react. There are situations that having the process lead makes all the sense in the world - for example trying to compete in the high core count arena, becomes difficult. However as companies that try to compete with Intel but until recently thought they would be in the very best case at process parity made interesting decisions (e.g. chiplets, multiple threads per core, accelerators etc etc). Intel should take an example and try to balance the architecture with what you can reasonably expect by the implementation. On the other hand, as previous commenters correctly pointed out, there are a lot of situations that bleeding edge is not as crucial and Intel should make the most of it, since it has quite an arsenal of products that do not really depend on having the most dense process (e.g. chipsets, wireless/wired networking, controllers, power electronics etc) and maybe these products could be a source of future growth more than what high-performance CPUs are now.

Many of these products (chipsets, controllers, power electronics), Intel is fabless on already. Intel should transition to becoming a fabless company, my opinion.

 
Many of these products (chipsets, controllers, power electronics), Intel is fabless on already. Intel should transition to becoming a fabless company, my opinion.

The only chips that I know of where Intel uses TSMC is from companies they have been purchased. Altera is the biggest one. FPGAs have a very long life and Altera was exclusive to TSMC (20nm and above) until the acquisition by Intel.

Given the pandemic, AMD's increasing market share and Intel 10nm ramping I do not see any 14nm capacity constraints.
 
There seems to be a generic malaise with US companies that depend on engineering as the core of their business model. Intel, Boeing, and General Electric are the most publicly troubled, but I am betting there are many others.
It has been predicted and warned, for decades, that older, experienced engineers will retire and there are no replacement for these trailblazers. It happened. The results are now being seen.
A correct diagnosis of the problem is required before a solution can be designed. The engineering ecosystem is not healthy. Millenial engineers have had to struggle to obtain entry level positions and haven’t gained the traction and experience to replace retiring engineers. There is no room for growth and no tolerance; there is no such thing as learning from mistakes (which is where the experience and wisdom comes from).
In the current environment, what used to be inspiring work is now deadening, with endless hassles.
These are the trends I connect to Intel failure to introduce a new node to volume production for 7 years, Boeing failure to produce a non-murderous 737, and GE’s general decline and unluck.
 
There seems to be a generic malaise with US companies that depend on engineering as the core of their business model. Intel, Boeing, and General Electric are the most publicly troubled, but I am betting there are many others.
It has been predicted and warned, for decades, that older, experienced engineers will retire and there are no replacement for these trailblazers. It happened. The results are now being seen.
A correct diagnosis of the problem is required before a solution can be designed. The engineering ecosystem is not healthy. Millenial engineers have had to struggle to obtain entry level positions and haven’t gained the traction and experience to replace retiring engineers. There is no room for growth and no tolerance; there is no such thing as learning from mistakes (which is where the experience and wisdom comes from).
In the current environment, what used to be inspiring work is now deadening, with endless hassles.
These are the trends I connect to Intel failure to introduce a new node to volume production for 7 years, Boeing failure to produce a non-murderous 737, and GE’s general decline and unluck.

I don't think the problem is that there aren't enough good engineers. There are lots. And it's not US companies that have this problem in general - for every Intel, Boeing, GE, you have an NVidia, SpaceX, and Google. The problem is there is very often a shift in focus that happens over time at large, mature companies from product development to marketing/sales. At a certain point, the path to higher profits lies not in improving your product, but developing your sales channels, optimizing pricing, market segmentation, ect. Companies start to prioritize these areas over product development because product development is more risky and expensive. The people who generate the most value for the business become the sales and marketing guys, and those are the people that end up running the companies. The pendulum swings too far and over time, this drives away many engineers, who want to develop new products, so they go to NVidia, SpaceX, Google, ect. It's all part of the cycle of creative destruction.
 
Based on Intel's most recent announcement, it seems like they are giving up on process leadership altogether. I'm just bumping to take my victory lap on this call ;)

Many of these products (chipsets, controllers, power electronics), Intel is fabless on already. Intel should transition to becoming a fabless company, my opinion.

 
There seems to be a generic malaise with US companies that depend on engineering as the core of their business model. Intel, Boeing, and General Electric are the most publicly troubled, but I am betting there are many others.
It has been predicted and warned, for decades, that older, experienced engineers will retire and there are no replacement for these trailblazers. It happened. The results are now being seen.
A correct diagnosis of the problem is required before a solution can be designed. The engineering ecosystem is not healthy. Millenial engineers have had to struggle to obtain entry level positions and haven’t gained the traction and experience to replace retiring engineers. There is no room for growth and no tolerance; there is no such thing as learning from mistakes (which is where the experience and wisdom comes from).
In the current environment, what used to be inspiring work is now deadening, with endless hassles.
These are the trends I connect to Intel failure to introduce a new node to volume production for 7 years, Boeing failure to produce a non-murderous 737, and GE’s general decline and unluck.

As a millennial engineer who is sick of hearing about the glory days of the industry, it's pretty clear that for a few decades now Wall St and it's insane free market ideology has replaced the generational transfer of technological expertise with outsourcing and arbitrage.

Like how Magnequench, the world's foremost processor of rare-earth metals , used to be in an Indiana company but now is a property of the CCP. It was a bold strategy, but by now there's no one with degrees in law or finance who can to alter the course of things.
 
Murthy is out....

"A company spokesperson declined to say whether he was fired but said Renduchintala's departure was the result of a reorganization that will split up the Technology, Systems Architecture and Client Group into separate teams."


 
I've referenced this Scotten's article on Twitter and a number of people seemed to have picked it up. It's a great article that although technical is easy for the lay person (like myself) to understand. Thank you Scotten & Dan for making this possible!
 
Back
Top