Question Susquehanna's second chat with Charlie Demerjian

[uzzi38]

https://www.reddit.com/r/AMD_Stock/comments/eczmzx

This is not a good picture Charlie's painting of whatever's left of Intel's roadmap.

For the record, this is also significantly worse than any expectations I ever had as well. Far, far worse.

 

[moinmoin]

Actually no. I don’t just listen to what the company tells me. I try to get corroboration from other sources. This is called “due diligence” in my line of work. But hey thanka for trying to interpret how I get my job done working at 10bln+ aum firm

You don't show much of that on this board, if at all.

 

[liahos1]

You don't show much of that on this board, if at all.

Don’t need to... I come here to try to find views (with actual backup) that are contrary to what I get from due diligence in the supply chain.

 

[moinmoin]

Don’t need to... I come here to try to find views (with actual backup) that are contrary to what I get from due diligence in the supply chain.

That's unfortunate since what you post here so far barely qualifies as views.

 

[liahos1]

Sorry not trying to impress anyone here. Trying the impress my investors.

 

[DrMrLordX]

Sorry not trying to impress anyone here. Trying the impress my investors.

There's this NostaSeronX guy . . . you should listen to him and pass along his advice. He is brilliant and has the best sources!

 

[jpiniero]

Don’t need to... I come here to try to find views (with actual backup) that are contrary to what I get from due diligence in the supply chain.

So why is the 14 nm shortage seemingly getting worse now, esp when volumes are way down?

 

[lobz]

So why is the 14 nm shortage seemingly getting worse now, esp when volumes are way down?

He must have taken the wrong quarterly Asian trip. If you know what I mean ;p

 

[NTMBK]

Intel are building out additional capacity on 14nm, at a time when they should be transitioning their entire CPU line over to the new, superior process. And bringing out yet another warmed over Skylake, even as Icelake is supposedly launched. Doesn't take a genius to figure that one out.

 

[liahos1]

All together now. This year intel increased wafer capacity about 25pct mix of 10/14nm. Next year it’s another 25pct wafer capacity primarily from 10nm.

the company has less than 2 weeks of finished goods pc cpu inventory.

they came into the q3/q4 with a build schedule based on a certain amount of demand expectation. A couple of things were off with their forecasts

Pc demand was a few hundred basis points higher than they anticipated. DCG demand for cascade was much higher for xcc than they anticipated.

coupled with no buffer inventory, they were short supply the market.

As it relates to yoy - if you all recall, intel has been running shortages since summer of last year. They announced a 1.5bln capex increase last year.

They were extremely short of inventory last year and as a result had to bleed off 1-1.5bln of buffer inventor.

effectively they were selling product in excess of their ability to produce it.

they didnt have that luxury this year

 

[liahos1]

Btw the Asia trip included meeting with inventec, asmedia, quanta, gigabyte, ASUS etcx

but that’s just one check. The other check is based on the folks that handle equipment logistics going into intel fabs and sell them consumables (like photovoltaic masks)

so while all you guys crap on me and try to tell me what’s going on, I’ll just watch and eat my popcorn

 

[jpiniero]

All together now. This year intel increased wafer capacity about 25pct mix of 10/14nm. Next year it’s another 25pct wafer capacity primarily from 10nm.

It was mostly 14 nm and clearly it wasn't enough. They increased the U dies from 2 to 4 to 6 cores and the desktop went from 4 to 6 to 8. And as you mentioned, due to Rome they are forced to sell XCC dies for the most part. But it seems so sudden that the shortage seemed to be easing and now it's getting worse again. Maybe it is Servers then... and if it is, Cooper's only going to make things worse. You can see why Cooper got delayed.

They did say they were adding 10 nm capacity in the future. I could see Intel thinking that they are so uncompetitive in servers now that they have to go 10 nm or bust in servers because they need the power savings to even have the conversation of being competitive with anyone using TSMC.

 

[lobz]

All together now. This year intel increased wafer capacity about 25pct mix of 10/14nm. Next year it’s another 25pct wafer capacity primarily from 10nm.

the company has less than 2 weeks of finished goods pc cpu inventory.

they came into the q3/q4 with a build schedule based on a certain amount of demand expectation. A couple of things were off with their forecasts

Pc demand was a few hundred basis points higher than they anticipated. DCG demand for cascade was much higher for xcc than they anticipated.

coupled with no buffer inventory, they were short supply the market.

As it relates to yoy - if you all recall, intel has been running shortages since summer of last year. They announced a 1.5bln capex increase last year.

They were extremely short of inventory last year and as a result had to bleed off 1-1.5bln of buffer inventor.

effectively they were selling product in excess of their ability to produce it.

they didnt have that luxury this year

OK Mr BigShot, explain to the forum again, how increasing wafer capacity has anything to do with the yields you achieve from those wafers, because while your investors may be impressed as hell, this particular information was left out of your comment somehow, which otherwise contained no information that was not already publicly available.

All you told us now is basically that intel has poured a crapload of money into 10nm wafers in 2019 with ridiculously poor results. Some smart investors they must be, whom you can convince to be impressed with what you've just told.

As for me being a wall street guy? I wish. But hey, at least I don't have this itching delusion of grandure that you've displayed in these last couple of comments. What actually convinced me now not to take seriously anything you say was this: you attacking me out of nowhere, and then 1-2 guys replying to you, what does all this mean to you? You see it as 'all you guys crapping on me'. Insecure much?

That's about as real as Charlie being paid for sharing what he knows. Like anybody would actually pay a guy to spread FUD who then goes ahead and puts the newest and juiciest informations behind a $100-1000 paywall.

 

[liahos1]

Hey Lobz - like I said - I thought you posted it and corrected. I was looking at the post 4am my time. I see it was uzzi and not you. If the prior post didn’t make that clear hopefully this one does.

yield comments come from piecing together commentary from the consumables folks, supply commentary from the Asia checks and intel management comments. There isn’t a 10nm supply issue.
israel lit up over the summer, Oregon lit up in October. Chandler lights up next month.

I can tell you yields are fine. They aren’t 14nm yields yet but it’s a economic node and getting better. In addition my checks Indicate 7nm is progressing quite well.

in terms of ad hominem attacks - it’s pretty evident on this forum. Whenever someone (like myself) has something positive to say about intel, 10nm etc your ilk comes out to slam it as if it’s a thought crime.

but like I said earlier - I don’t really care. I’m eating my popcorn watching you guys do it. It doesn’t matter because my color is good (been investing in semis for over 10 years) and I’m just waiting for this to play through.
There is gonna be a lot of salt coming out of these boards in 2020-2021.

 

[liahos1]

It was mostly 14 nm and clearly it wasn't enough. They increased the U dies from 2 to 4 to 6 cores and the desktop went from 4 to 6 to 8. And as you mentioned, due to Rome they are forced to sell XCC dies for the most part. But it seems so sudden that the shortage seemed to be easing and now it's getting worse again. Maybe it is Servers then... and if it is, Cooper's only going to make things worse. You can see why Cooper got delayed.

They did say they were adding 10 nm capacity in the future. I could see Intel thinking that they are so uncompetitive in servers now that they have to go 10 nm or bust in servers because they need the power savings to even have the conversation of being competitive with anyone using TSMC.

They aren’t forced to do anything. Customers are up selecting higher performance parts. Asps are going up and have been going up across their stack since even zen 1 launch.
For all the intelligence in this board it’s baffling you guys seem to simply ignore that.

intel grew its DCG revenue 1.4bln sequentially. Amd grew epyc revs 80mln dollars. Thats just reality. The Asia checks I got indicate amd should continue picking up share but the folks weren’t that optimistic on the 10pct target by mid year.
Don’t shoot the messenger

 

[jpiniero]

They aren’t forced to do anything. Customers are up selecting higher performance parts. Asps are going up and have been going up across their stack since even zen 1 launch.

Because customers who were shrewd enough got Intel to give them Plat for slightly more than the typical Gold or even Silver price. Kind of baffled that was enough to ensure they didn't defect to AMD. Maybe the real pain will be when the next validation cycle occurs.

 

[RetroZombie]

the company has less than 2 weeks of finished goods pc cpu inventory.

That is a part that I don't understand, I have an friend that works in the distribution channel and he was surprised that for the Q4/2019 and Q1/2020 the huge amount of new laptops that still comes with intel 7th gen cpus.

If intel doesn't have capacity to manufacture enough stuff and it obviously runs out of stock, but then instead of the 'new' 8th, 9th and 10th gen they still sell more than 30% parts with 7th gen cpus. Something clearly doesn't add up.

 

[Arkaign]

Do you have any information that makes any of this false ? Not just feelings, anything in writing ? A link ? While they may seem anti-Intel, I see nothing to speak of that goes against everything we have heard. And SEEN.

Yeah, both can and are true in this case.

Charlie and his site are notoriously, zealously, religiously anti-Intel to an extreme level.

And, Intel has been fumbling for quite some time now, with no end in the near future at the very least.

 

[Glo.]

To everybody who has some emotional investment, anything that states the obvious, in specific language, like Charlie uses will appear as he is hating Intel.

But he is not hating them. He just points blatant incompetency that is inside the company, lately. You may not like it, but Charlie was talking about 10 nm woes since 2015. Like they have turned untrue, right?

He was making stuff up because he hated Intel, right?

Give Charlie credit, when its due. He was right way more often on Intel topic, than not right, at all .

Oh, and last thing. Charlie on topic of Intel has way more credibility than some random forum user that claims to have spoken with Asian Suppliers .

 

[Arkaign]

No lol, I've been following him way longer than that, and know his rantings very well. He's correct lately because Intel has been a dumpster fire for a while now, but he's 100% of the time always been an Intel hating troll. I don't listen to trolls, there are others who are rampant AMD haters, Nvidia haters, etc. It makes people emotional and unreliable. Confirmation bias when it suits them doesn't make them reliable or unbiased.

 

[lobz]

No lol, I've been following him way longer than that, and know his rantings very well. He's correct lately because Intel has been a dumpster fire for a while now, but he's 100% of the time always been an Intel hating troll. I don't listen to trolls, there are others who are rampant AMD haters, Nvidia haters, etc. It makes people emotional and unreliable. Confirmation bias when it suits them doesn't make them reliable or unbiased.

What makes you think we haven't been following him as long as you or even longer? Just curious how did you become the leading authority on deciding if he's a troll or not. How is it a confirmation bias when one of the only things he didn't foresee was how intel's marketing is going to spin the dreadful and embarrassing death of their original 10nm plans.

 

[LikeLinus]

It's well known he and his site is very anti-intel. You can Google and find a bunch of threads discussing it. It's not some big secret. If you want to believe he is not anti-intel and has no bias, more power to you.

 

[liahos1]

He’s also said on Twitter he doesn’t do wall st. hosted calls. He’s full of it. When I called him out on Twitter after he did his first Susq call he blocked me. He’s not a journalist. He’s a rabble rouser

 

[lobz]

He’s also said on Twitter he doesn’t do wall st. hosted calls. He’s full of it. When I called him out on Twitter after he did his first Susq call he blocked me. He’s not a journalist. He’s a rabble rouser

Write this down here a hundred more times in case abody missed it the first couple of times and then we will start to believe you.
What happened again when you called him out on twitter?
Other than feelings got hurt, I mean.

 

[lobz]

It's well known he and his site is very anti-intel. You can Google and find a bunch of threads discussing it. It's not some big secret. If you want to believe he is not anti-intel and has no bias, more power to you.

If that was an answer to me, first read the post I replied to. Bias and confirmation bias are 2 different things. I absolutely believe he has a bias against intel, as they are one the most dishonest tech companies out there. Confirmation bias? You could make a case of that against anybody in the world...

 

[Arkaign]

What use is obviously biased material though? I guess it depends on your priorities.

It's why I started following AnandTech in the first place, Anand Shimpi was incredibly professional and even handed. He called it exactly the way it was in an objective fashion. Or today like with Tech Jesus.

If company A does 7 dumb things, and 3 good ones, and company B does 8 good things, and 2 stupid ones, then I want someone that tells it honestly, without a vendetta or a case of the shills.