AMD executive says Intel Foundry Services is destined to fail

[Alfonso Maruccia]

A hot potato: Intel CEO Pat Gelsinger's decision to turn the Santa Clara corporation into a chip manufacturing foundry and open it to orders from third-party companies hasn't impressed AMD. Team Red explains that it's the wrong choice which will ultimately bring Intel to its downfall.

Intel's Foundry Services (IFS) offers "world-class wafer foundry" services to chip companies that don't have a manufacturing arm. Chipzilla is opening its advanced manufacturing sites, advanced packaging technology and development services even to competing companies, as its CEO Pat Gelsinger is trying to open new revenue sources in an increasingly competitive and shrinking PC market.

Will Intel succeed in this seismic shift in its traditional business structure? "Of course not," said Darren Grasby. An AMD VP for strategic partnerships and president of AMD EMEA, Grasby slammed Intel's decision during the Canalys EMEA Forum 2023, stating that the company's new path to revenue growth is the wrong one.

AMD made the opposite journey years ago, Grasby said, abandoning its own chip manufacturing fabs and choosing a totally "fabless" route. The x86 company was thus able to invest a lot more money into the roadmap of its processor designs, which is ultimately bringing AMD's "leading edge technology" and successful Ryzen products to the market today.

Intel has always been a chip manufacturing company, at least when it comes to its powerful x86 CPUs. Under Gelsinger's tenure, the corporation adopted a more open approach for newer product lines like Arc GPUs, which are made by Taiwanese foundry TSMC. Intel is also making significant investments in new production facilities for the IFS initiative, with new manufacturing plants being built in the US (Arizona, Ohio, New Mexico) and Europe (Germany).

Speaking about IFS, Gelsinger recently said that he and the company would receive a "passing grade" two and a half years into the journey. If the foundry initiative is successful, Gelsinger stated, Intel will have to "accept business" opportunities from its competitors as well, which means that "made by Intel" chips would bring in revenue from the likes of Nvidia, Arm and even AMD.

Market analysts recently speculated that Intel could soon split in two, with a R&D arm becoming customer of a newly created (and essentially independent) chip foundry business. According to IDC senior analyst Andy Buss, having more choices among chip foundries is "good" for the entire chip market. TSMC is running out of manufacturing space, and AMD must compete with the likes of Nvidia, Apple and other prominent technology corporations to make its chips.

Permalink to story.

https://www.techspot.com/news/100424-amd-executive-intel-foundry-services-destined-fail.html

 

[WhiteLeaff]

For the first time in a long time, I disagree with AMD. Fabs are a good business for Intel, especially in mature nodes that can serve to support local manufacturing of EVs.

 

[kira setsu]

When everything needs a chip it would be wise to yknow, be able to have enough, which is still a problem so focusing on that part is very very smart to do.

AMD is just asking to always be someones number 2 isn't it.

 

[wiyosaya]

With Intel's widely publicized foibles around its attempts at what was it? 10nm, I am not so sure that Intel is wisely endeavoring to enter the foundry business. Whether AMD is right or Intel will succeed as a foundry company remains to be seen. My bet, however, is if they do succeed, they will be very selective in whom they choose to make chips for. After all, they will likely exclude making chips for their competitors.

 

[VcoreLeone]

Care to explain why ?

 

[SalmanIsHere]

Amd has been really cocky recently.
This is a sign of weakness in their own internal plans.
12th gen destroyed the superioty amd had with zen.
They are 10 years behind Nvidia.
Lisa even a year ago said AI has not future.
Things are not looking good for amd's future.
Competition from all fronts.

 

[kiwigraeme]

With Intel's widely publicized foibles around its attempts at what was it? 10nm, I am not so sure that Intel is wisely endeavoring to enter the foundry business. Whether AMD is right or Intel will succeed as a foundry company remains to be seen. My bet, however, is if they do succeed, they will be very selective in whom they choose to make chips for. After all, they will likely exclude making chips for their competitors.

With other companies - won't be a problem for AMD
however - If Intel has extra compacity - I think they quite of have to take AMD business as using govt subsidies
. Plus it's probably good for their bottom line - If they have a design problem - that's not AMDs fault.

LG , Samsung sell to each other , Samsung sells to Apple - even at height of legal battles
Business is business

Intel foundries will need customers - having a competitor is a huge coup

I mean how many customers for high end are there - Don't let Nvidia use it for GPUs , AI as Intel makes GPU

Nvidia will try and move into SOC space as well to a greater degree when such SOCs are enough for powerful handheld devices

 

[Tom Yum]

I think AMD's comments are more about the current approach Intel is taking to IFS, rather than a statement that the industry doesn't needs more manufacturing competition (which it desperately does).

Intel's current approach with IFS is to have their cake and eat it too. They want the economies of scale provided by fabbing third parties products, but also want to continue competing with the very same companies that might use IFS to fab their products.

Right now Intel has had very little success getting other companies to sign up for IFS services. $232 million in revenue in the last quarter which is peanuts compared to the tens of billions being plowed into it. A large part of that is the simple fact Intel competes with a large number of potential companies that might use IFS. I'm not saying Intel would steal competitor IP (that would be suicide), but they would still get a close look at the competitors product planning and derive an advantage from that. Understandably, that makes companies hesitant to deal with IFS.

The only solution would be Intel completely separating (as in, two separate companies, with no ownership share) into IFS and Intel Design to eliminate the conflict of interest. Intel Design would then contract with IFS to manufacture Intel design products, but at market rates. That is not Pat's plan though, and I think AMD's comments speak to that issue.

 

[toooooot]

Regardless of what AMD exec said,
I am very curious to see how these tow
companies and their different routes will turn out
in the next 5 - 10 years

 

[sreams]

Amd has been really cocky recently.
This is a sign of weakness in their own internal plans.
12th gen destroyed the superioty amd had with zen.
They are 10 years behind Nvidia.
Lisa even a year ago said AI has not future.
Things are not looking good for amd's future.
Competition from all fronts.

10 years behind NVidia? Are you actually suggesting that current gen AMD GPUs are at the same level as the GTX 7xx series from 2013?

 

[Bobbydpue]

10 years behind NVidia? Are you actually suggesting that current gen AMD GPUs are at the same level as the GTX 7xx series from 2013?

I was going to say the same thing. What the OP said was just way out of line and it's not even close to being true. At worst they are a single or two generations behind and that's mostly for their upscaling tech. If AMD didn't raise their prices along with NVIDIA, they would be a leader in GPU sales, but just like I called it years ago as soon as AMD can raise prices, they will do it and they did.

 

[redgarl]

What this executive from AMD is saying is that AMD current success is attributed to focusing on what they were the best at, which was chip design, and it worked.

IFS will fail miserably because Intel cannot focus on what they were the best at, which is not process node or manufacturing. It is actually one of their worst attribute.

They diversified their portfolio of IPs so much that they are good at nothing anymore. Memory, 5G, IoT, GPU, Network... name them, all failed ventures funded on their cashcow, CPUs for mobile and servers.

Anyone thinking Intel can catchup with TSMC are out of their mind. CoWoS is the future and Intel is like 5-10 years behind TSMC on that strategy. You will not get AMD or Nvidia chip business if you cannot provide CoWoS.

 

[redgarl]

Amd has been really cocky recently.
This is a sign of weakness in their own internal plans.
12th gen destroyed the superioty amd had with zen.
They are 10 years behind Nvidia.
Lisa even a year ago said AI has not future.
Things are not looking good for amd's future.
Competition from all fronts.

If AMD was 10 years behind, then tell me why AMD is beating Nvidia in all dGPU segment EXCEPT the 1600$ WHALE tier?

Also, you probably missed that AMD compute GPUs are practically on par with Nvidia GPUs in AI in an OPEN AI environment, and pretty much better when it comes to compute alone.

https://www.techspot.com/news/99271-amd-ai-chips-punch-above-their-weight-mosaicml.html

 

[maxxcool7421]

This is a pure tax dodge .. nothing to see here

 

[Watzupken]

There are no perfect approaches. Each have its own merits. AMD is not wrong based on their own experience, and being tied to Global Foundaries almost always put them in a fab disadvantage. The situation is the same for Samsung now as they are stuck on their own fab to produce Exynos which is making their SOC a lot less performant than it should be. Other point to note is that a lot of big players have expanded their fabs significantly over the last 3 years. Having another big player open their fab door may not mean the fabs will automatically run at 100% capacity and selling like hot cakes. Hence it remains to be seen if Intel can find success here.

 

[Bobbydpue]

If AMD was 10 years behind, then tell me why AMD is beating Nvidia in all dGPU segment EXCEPT the 1600$ WHALE tier?

Also, you probably missed that AMD compute GPUs are practically on par with Nvidia GPUs in AI in an OPEN AI environment, and pretty much better when it comes to compute alone.

https://www.techspot.com/news/99271-amd-ai-chips-punch-above-their-weight-mosaicml.html

Doesn't NVIDIA have a majority share in the dGPU market? After a quick check in Q1 AMD had 12% market share while NVIDIA had 84%. Did AMD suddenly sell piles of GPUs between Q1 and this quarter?

I'm not fanboying here, just trying to figure out where you are pulling data from to form your opinion since objective data doesn't support your claim.

 

[RaXelliX]

Doesn't NVIDIA have a majority share in the dGPU market? After a quick check in Q1 AMD had 12% market share while NVIDIA had 84%. Did AMD suddenly sell piles of GPUs between Q1 and this quarter?

I'm not fanboying here, just trying to figure out where you are pulling data from to form your opinion since objective data doesn't support your claim.

He probably meant sales, not market share.

You have to remember that Nvidia's big market share does not consist of 40 series owners. They have a lot of users who use older cards from 10 and 20 series.