Question AMD to double EPYC production rate in 2023

bsp2020

Member
Dec 29, 2015
98
104
106

I’m not sure if I’m reading it correctly. Forrest Norrod says “And just to be clear, we are planning for doubling year-on-year over time.” in response to the question of why AMD is still having issues with sourcing advanced substrates. Am I to understand it as AMD planning to double EPYC production rate in 2023? What do you think?
 

moinmoin

Diamond Member
Jun 1, 2017
4,046
6,086
136
I think he's referring to revenue which at an increasing ASP doesn't equal quantity. But in general AMD keeps indicating that with Zen 4 on N5 its prepared to ship a significantly higher quantity of chips. Looks like they are confident to have a handle on the previous gens' supply bottlenecks now.

Edit: I see in the interview the topic is actually explicitly about quantity. So I'd take him at his word.

Norrod also repeated what AMD has been saying for quite some time, that TSMC is not the bottleneck but substrates are:

"The principal gate for us is not wafers. Particularly for these Epyc chips, it’s advanced substrates. And there’s just a long lead time to build up the factories and increase capacity for those substrates. We have made major investments and I think we are ramping that capacity at a very steep but prudent rate."

As they have been saying that for quite some time already it's not farfetched that the time for resulting improvements is now.
 
Last edited:

eek2121

Platinum Member
Aug 2, 2005
2,212
2,867
136
I think he's referring to revenue which at an increasing ASP doesn't equal quantity. But in general AMD keeps indicating that with Zen 4 on N5 its prepared to ship a significantly higher quantity of chips. Looks like they are confident to have a handle on the previous gens' supply bottlenecks now.

Edit: I see in the interview the topic is actually explicitly about quantity. So I'd take him at his word.

Norrod also repeated what AMD has been saying for quite some time, that TSMC is not the bottleneck but substrates are:

"The principal gate for us is not wafers. Particularly for these Epyc chips, it’s advanced substrates. And there’s just a long lead time to build up the factories and increase capacity for those substrates. We have made major investments and I think we are ramping that capacity at a very steep but prudent rate."

As they have been saying that for quite some time already it's not farfetched that the time for resulting improvements is now.
Unsure where you get that. he is clearly talking about production. Read the article.
 

maddie

Diamond Member
Jul 18, 2010
4,337
3,877
136
I think he's referring to revenue which at an increasing ASP doesn't equal quantity. But in general AMD keeps indicating that with Zen 4 on N5 its prepared to ship a significantly higher quantity of chips. Looks like they are confident to have a handle on the previous gens' supply bottlenecks now.

Edit: I see in the interview the topic is actually explicitly about quantity. So I'd take him at his word.

Norrod also repeated what AMD has been saying for quite some time, that TSMC is not the bottleneck but substrates are:

"The principal gate for us is not wafers. Particularly for these Epyc chips, it’s advanced substrates. And there’s just a long lead time to build up the factories and increase capacity for those substrates. We have made major investments and I think we are ramping that capacity at a very steep but prudent rate."

As they have been saying that for quite some time already it's not farfetched that the time for resulting improvements is now.
I know this is an AMD thread, but how will this affect Intel revenue and margins?
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
23,817
12,913
136
I know this is an AMD thread, but how will this affect Intel revenue and margins?
If they sell twice as many EPYC, you don't think that might affect Intel sales ? In the past, AMD has lost a lot of sales because they did not have the product to sell.
 

maddie

Diamond Member
Jul 18, 2010
4,337
3,877
136
If they sell twice as many EPYC, you don't think that might affect Intel sales ? In the past, AMD has lost a lot of sales because they did not have the product to sell.
Yes I do, but to what extent, is the question.
 

moinmoin

Diamond Member
Jun 1, 2017
4,046
6,086
136
Unsure where you get that. he is clearly talking about production. Read the article.
I did and corrected myself, read the edit you even managed to quote. Want me to delete the early take so you can find the edit?

Yes I do, but to what extent, is the question.
So far AMD was essentially able to sell everything so if they manage to keep doing that the increase in production will directly translate in loss of sales and shares by Intel. This is exacerbated by the current macro economic environment where Intel already indicated that its products are less requested whereas AMD's products are so far still requested more than supply is available.

Essentially two parts are important:
  • Did Intel prepare for the current slow economy (as it should)? If yes, it should be able to handle AMD's production increase as part of it.
  • Will demand for AMD products finally fall below supply (as it is expected to happen eventually)? If it does AMD, now being used to be able to sell everything they can produce, may itself be ill-prepared.
 

LightningZ71

Golden Member
Mar 10, 2017
1,490
1,698
136
I believe that, what is still working in AMD's favor, the economy is still shifting to the cloud model across the business landscape, and that shift is largely independent of the economy overall. Where as the demand for general computing resources may shrink a little, the demand from businesses retiring local servers and in-house data centers to move to cloud based solutions as a cost saving measure will actually increase as the global economy circles the toilet bowl. Companies will want to reduce in-house costs, including overhead from their own staff, and shift the security burden, at least to some extent, onto third parties that they can hold financially liable in court. Add in the fact that older cloud server hardware has an expected service life, and a lot of "first-wave" cloud gear is reaching EOL and needs to be replaced.

The server market may flatten a little overall, Intel still can't ship the volume on their leading edge product/node combination like they used to, and AMD is adding volume to meet market demand that competitors can't meet due to their own undesirable products and limited volume (Intel) or due to technical limitations (ARM products) with respect to compatibility with existing software bases. The landscape is still shifting as cloud vendors are still adding capacity overall and will eventually reach market saturation as well as more and more products becoming platform agnostic due to a shift to a thin/rich client model based around web delivery that doesn't care about it's back end.
 

Abwx

Diamond Member
Apr 2, 2011
9,816
2,252
136
What is of most importance currently is the perf/watt given the abruptly rising electricity costs, guess that those who did choose Intel s aging plateform instead of AMD s 7nm two years ago are now looking at the revised TCOs with awe...
 
Last edited:

Joe NYC

Senior member
Jun 26, 2021
934
1,075
96
I know this is an AMD thread, but how will this affect Intel revenue and margins?
Intel Datacenter is barely above break even already. Intel is desperately trying to hold on to the market share, by giving the server chips almost at a loss:

1664913073741.png

I think what can happen though is with the decline in Client revenue, Intel may be running into low fab utilization, which has high fixed costs.

And Intel already signed TSMC contracts, and will be obligated to take deliveries from TSMC, even while Intel's own fab will have spare capacity.

So not a good confluence of events for Intel...
 

moinmoin

Diamond Member
Jun 1, 2017
4,046
6,086
136
Intel is desperately trying to hold on to the market share, by giving the server chips almost at a loss:
While the numbers are undeniable bad, operating income/margin do include investments and inventory which are not necessarily straight losses. Gross income/margin would be more interesting there.
 
  • Like
Reactions: coercitiv

naad

Member
May 31, 2022
62
175
66
That's what happens when you introduce a new product into the mix that uses a new node, near overnight 60%+ increase in volume, you don't stop selling N7 SKU and you won't stop selling them even when N3 hits. Milan is still much, much cheaper to make than Sapphire Rapids, lower TCO and has quite a few advantages with V-cache SKU.

Most importantly it's "tried and tested" like Skylake-SP and Cascade Lake were. Milan is what small and medium business not using AWS/Azure/etc will use going forward.
 

ASK THE COMMUNITY