News Intel CEO Pat Gelsinger Declares 'AMD's Lead Is Over' After Alder Lake, Sapphire Rapids

[Kj1985]

The question was if they can buy enough to limit the amount of wafers that intel could buy, not if they can buy enough to make a profit for themselves.

If "the leftovers" is the large majority of wafers then it doesn't make any sense.

No, that was never the question I was answering. The original post I was responding to was saying that TSMC will ditch AMD for Intel. And their proof was a link to a rumor.
AMD will never buy up fab capacity to shut out Intel, that does not make business sense. They will only buy what they need to produce their products.

 

[Zerk2012]

Well that was a short run for AMD to bad it's over. If the CEO of Intel said it then it has to be true.

 

[$h0nuff]

Most proficient CEOs answer questions like this because there are shareholders, board members, customers, etc who are reading/watching. Their job is to make/keep all of these groups happy. AMD Lisa and many others have answered questions in the same manner. Given the nature of the business and tested yields, with release a month out, I surmise the offerings will be competitive.

Most older members on these types of boards have used Intel, AMD; different video cards and peripherals, as leads and performance features change hands over time. Taking a fanboi stance based on your personal purchasing preferences doesn't offer much to the conversation IMHO.

3rd party foundry time has more to do with advance scheduling than it does operating expense figures. TSMC and Samsung don't look at revenue statements and play favorites. Nor do they push one run out of the way for another because company x is "preferred". They want their equipment running consistently and with advanced knowledge of what's next.

On the same token, the companies using these foundries need to have a design ready to go. They don't barter time and then design...quite the other way around. Smaller (scheduled) engineering design and test yield runs followed by a larger batch schedule pending sucess.. Same process if it's an Intel or AMD (Global Foundries) in-house product run.

 

[spongiemaster]

No, that was never the question I was answering. The original post I was responding to was saying that TSMC will ditch AMD for Intel.

Then you were responding to no one, because no one has said that in this thread.

 

[RodroX]

Well that was a short run for AMD to bad it's over. If the CEO of Intel said it then it has to be true.

Amen

 

[Kj1985]

Then you were responding to no one, because no one has said that in this thread.

From you first post:
'Intel is about to have a process advantage vs AMD because of the same company you just said is going to take them down, TSMC. Intel has more money than AMD, they now have priority over them for leading edge node allocations as a result. Rumors are that Intel will be using TSMC's 3nm for some Xeon sku's, so the market could look very different by the end of 2022.'
You said it in the second sentence. 🤣

 

[spongiemaster]

From you first post:
'Intel is about to have a process advantage vs AMD because of the same company you just said is going to take them down, TSMC. Intel has more money than AMD, they now have priority over them for leading edge node allocations as a result. Rumors are that Intel will be using TSMC's 3nm for some Xeon sku's, so the market could look very different by the end of 2022.'
You said it in the second sentence. 🤣

Uhh, yea. No where in that quote do I say or even imply that TSMC is going to stop working with AMD. Nothing left to discuss with you.

 

[TerryLaze]

Well that was a short run for AMD to bad it's over. If the CEO of Intel said it then it has to be true.

Intel is super happy with AMD doing well in servers and do you know why?
Because intel only cares about making money.
In 2016, before ryzen released intel made 17.2 bil from the data center group.
In 2020 when AMD stole 10% market share from intel intel made 26.1 bil from DCG
So AMD doing well made intel more money then it did AMD.
The only reason that intel talks about AMD is PR, it's just good sportsmanship to acknowledge the competitions accomplishments.

 

[Zerk2012]

Intel is super happy with AMD doing well in servers and do you know why?
Because intel only cares about making money.
In 2016, before ryzen released intel made 17.2 bil from the data center group.
In 2020 when AMD stole 10% market share from intel intel made 26.1 bil from DCG
So AMD doing well made intel more money then it did AMD.
The only reason that intel talks about AMD is PR, it's just good sportsmanship to acknowledge the competitions accomplishments.

All companies only care about making money it's why their in business.

 

[JayNor]

Raja stated that the fab decision was made at the start of design ... so probably 2018 for Xe-HPG that are now using N6. AMD could have reserved N6 back then, but didn't, and so are still on N7/DUV for Milan.

Intel is already using TSM N5 on Xe-HPC compute tiles, which Raja has also said was an early design decision ... my guess is, probably because Intel had no confidence that their EUV processing would be ready.

Intel is apparently building something with TSM N3 starting in mid 2022. The guesses about it being for Meteor Lake's GPU tile make the most sense to me.

If Intel builds any x86 chips at TSM in the near term, I'd guess it would be tiles of a Gracemont successor, which could be used for their 5G basestation family of chips. However, Intel has said it will be opening their x86 IP for IFS customers and will convert to using design tools compatible with the external foundries ... so, it sounds like eventually even their own server products could be manufactured at TSM.

 

[waltc3]

Poor Gelsinger...constantly comparing what Intel will be shipping to what AMD has been shipping for a long time... Myself, I want to see if Intel can keep up with Zen 4. Amazing how Gelsinger thinks AMD has shipped all it can...

Please explain to me the point of making statements like this while not announcing a ship date for either architecture?...Think you can actually do that some day, Intel? It's really getting old.

 

[zodiacfml]

you can't beat AMD with that if you're going to charge an arm and a leg. additionally, AMD's v-cache appears to be more potent than what Alder Lake can do for gaming.

 

[korekan]

This is like saying to make 9 women pregnant to get one kid in one moth instead of in 9 months.
Things that are not possible are not possible.

cpu can be designed, its a mechanism. if designed right starting from the assembly, execution, probe its possible.
it might not x86 or arm anymore. if we think about current tech, yes its not currently possible.
quantum computing is more likely like it.

 

[watzupken]

I think Intel failed to acknowledge that ARM is their true competitor, particularly in the enterprise/ data center market which is the most profitable. While AMD have been successful in chipping away market share from Intel over the past few years, they are still a a small competitor as compared to Intel. And unless Intel is willing to drop their profit margin, chances of them recovering all the lost market share is very low. Apple basically showed the world, the cost of using Intel chips vs their custom chip when they cut selling price of their MacBook Air/Pro, Mac and Mac mini.
Anyway, it doesn't matter who produces the best chip to me. As long as I get a fast chip at a good price, that is all I look out for. Just like I bought a i7 10700K recently as it was 60% of the price of a 5800X (MSRP price).

 

[watzupken]

Check the tier system again. It goes like this: You have the most money? Pick what you want.

It's been widely reported that Apple and Intel have squeezed out AMD from initial access to TSMC's 3nm. Stay up to date.

It is probably true that AMD don't have initial access to 3nm, but I don't think that is the strategy of AMD anyway, at least from what I see. They use the the next best fab that's available to avoid paying top dollars for it. Intel may have the cash to soak up whatever spare capacity that Apple is not using, but though Intel may be a big company, they are no match for Apple. So at the end of the day, competing with Apple for cutting edge node is like throwing away money. Intel on the other hand is more likely to use the cutting edge 3nm for their enterprise/DC chips since it is going to cost a lot and makes sense to use it for high margin products. Chances of consumer CPUs getting 3nm chip initially is very slim.

 

[watzupken]

you can't beat AMD with that if you're going to charge an arm and a leg. additionally, AMD's v-cache appears to be more potent than what Alder Lake can do for gaming.

I don't know about Intel's current CEO, but AMD's rise to fame started with humble pricing for double the cores. The cores ain't as good as Intel's at the start, particularly at single core performance, but it is still a good value for most looking for a good all rounded chip. My point is, as long as AMD is still producing very competitively priced products with decent performance, chances of AMD falling back is quite slim, unless Intel is willing to drop prices to compete.

 

[VforV]

Check the tier system again. It goes like this: You have the most money? Pick what you want.

It's been widely reported that Apple and Intel have squeezed out AMD from initial access to TSMC's 3nm. Stay up to date.

Others tried to tell you that although intel gets access to 3nm before AMD, it will 1) be the first iteration of 3nm and 2) in short supply (left from Apple allocation). The 1st iteration of a new node works well for Apple because they use it for mobile, which is smaller and less "pretentious" to say, but for CPUs is different story.
AMD will get 3nm when it will be refined and in bigger quantity available, so basically Apple as usual and even more now intel will be guinea pigs for the new nodes.
AMD has everything to win from this, intel wins just some PR for unknowledgeable people and most likely the products made on 3nm will be for server, not consumers, because of those low yields, low supply and high defect ratio the 1st iteration and all those other issues that come with it...

And on of the most important things that some people ignore is that TSMC and AMD are now partners for many years and they do share secrets to each other, but intel and especially with the IDM 2.0 plan will be a direct competitor to TSMC now.... so even if they give intel access to 3nm, they will have a much more limited and non-important relation than TSMC has with AMD.

 

[TerryLaze]

you can't beat AMD with that if you're going to charge an arm and a leg. additionally, AMD's v-cache appears to be more potent than what Alder Lake can do for gaming.

So you think that the only option for cooperate is to wait until AMD manages to get enough chips together?!
I don't think that that's how big business works, if they need CPUs now they are going to buy CPUs now and not wait for a time if any where they can find AMD CPUs.
For AMD to become a danger for intel's sales they would have to be able to produce at least the same volume as intel does.
Intel can charge anything they want and unless people can switch to arm or GPUs, which if they could they would have already done that years ago, people have no other choice than to buy intel because they are the only ones that can produce enough CPUs for the whole industry.

 

[dalek1234]

Desperate claims, as it is expected from Intel

 

[Deleted member 14196]

I still would never buy intel

 

[hannibal]

The 3D cache is going to put the hurt on them if Intel can't make it happen for them soon. Check HUB's core scaling investigation where they basically find that cache is the only deciding factor gen-over-gen for Intel!

Regards.

Maybe, but 3d cache is very expensive and 3d cache versions come much above normal zen3 chips that don´t have 3d cache. It is hard to say yet if 3d cache is the worth of extra price it will have compared to non cache versions.
It definitely gives a boost! So target will be higher. But so will be the price and that is important to the customers.
Interesting to see how much more expensive 3d cache versions are gonna be. It seems that alder lake by itself is significant jump in price allso...

 

[-Fran-]

Maybe, but 3d cache is very expensive and 3d cache versions come much above normal zen3 chips that don´t have 3d cache. It is hard to say yet if 3d cache is the worth of extra price it will have compared to non cache versions.
It definitely gives a boost! So target will be higher. But so will be the price and that is important to the customers.
Interesting to see how much more expensive 3d cache versions are gonna be. It seems that alder lake by itself is significant jump in price allso...

Your last line is basically what summarizes my reply to you: Intel is not going to be cheap with Alder Lake and upcoming CPUs anyway. AMD hasn't been cheap so far either, so I wouldn't be surprised, if they can overtake Alder Lake, to go even higher than now; if they manage to keep the costs around the same, I'd be honestly surprised. Then it becomes a question of how many people will actually be willing to upgrade to that price range given the extra performance. Then you look at individual CPUs pricing and decide then, I guess?

On the other hand, remember that CPU cost alone, particularly for this new generation, won't be the only factor to consider: DDR5, PCIe5 devices and new motherboards won't come cheap as a whole. Cooling if you're not lucky enough to carry over the cooling solution you may already have. A new build is going to start at roughly $400 for Alder Lake.

Regards.

 

[DanGer71]

The only thing TSMC has loyalty to is money, not AMD. You might want to try and read the news before posting dumb stuff like this.

https://www.techspot.com/news/90780-intel-has-reportedly-secured-majority-tsmc-3nm-production.html

Intel is about to have a process advantage vs AMD because of the same company you just said is going to take them down, TSMC. Intel has more money than AMD, they now have priority over them for leading edge node allocations as a result. Rumors are that Intel will be using TSMC's 3nm for some Xeon sku's, so the market could look very different by the end of 2022.

If TSMC is only intersted in money they would be dumb to allow AMD to get injured especially for a short term gain when they know Intel is building out process just to go back to their own as well as to become a TSMC competitor by outsourcing.

 

[DanGer71]

I expect Intel to win in some workloads with ADL vs Zen3. Zen3 with vcache is going to hit soon after and should close most benchmarks especially gaming ones. With regard to Sapphire, it will also close the gap with Milian but again the Milan with v cache should even it up. What Intel has to be concerned about is Zen4 with a supposed 29% IPC increase especially in the server space with Genoa. It will only be a quarter or two behind Sapphire so Sapphire's lead (assuming it has one at the time) will be short lived. Things will improve for Intel but competition will remain intense. I still give the edge to AMD over the next 2 years especially in server.

 

[spongiemaster]

Others tried to tell you that although intel gets access to 3nm before AMD, it will 1) be the first iteration of 3nm and 2) in short supply (left from Apple allocation). The 1st iteration of a new node works well for Apple because they use it for mobile, which is smaller and less "pretentious" to say, but for CPUs is different story.
AMD will get 3nm when it will be refined and in bigger quantity available, so basically Apple as usual and even more now intel will be guinea pigs for the new nodes.
AMD has everything to win from this, intel wins just some PR for unknowledgeable people and most likely the products made on 3nm will be for server, not consumers, because of those low yields, low supply and high defect ratio the 1st iteration and all those other issues that come with it...

I can try to tell you the world is flat, and that wouldn't make me right. Only one person has tried to argue my point, and they're wrong.

You need to be careful. No one here wants to see you hurt yourself straining so hard to spin this fab situation with TSMC into a positive for AMD.

And on of the most important things that some people ignore is that TSMC and AMD are now partners for many years and they do share secrets to each other, but intel and especially with the IDM 2.0 plan will be a direct competitor to TSMC now.... so even if they give intel access to 3nm, they will have a much more limited and non-important relation than TSMC has with AMD.

See below.

If TSMC is only intersted in money they would be dumb to allow AMD to get injured especially for a short term gain when they know Intel is building out process just to go back to their own as well as to become a TSMC competitor by outsourcing.

You're grossly overestimating AMD's importance to TSMC. Last year, by revenue, AMD was TSMC's sixth largest customer at 7.3%. Apple is #1 at a significantly higher 24.2%. Who was seventh? Intel, 1.3% behind AMD. Intel has been a major customer of TSMC for years. Almost as big as AMD. It's not in the news because until now, Intel hasn't been using TSMC for any of their flagship products. TSMC would be just fine without AMD as a customer. AMD needs TSMC more than the reverse. Where is AMD going without TSMC? Not back to Global Foundries. Not Intel. They're only option is Samsung. Ask Nvidia how that's working out for them.