AMD to outsource/sell Fabs (confirmed)

[cadaveca]

Why would anyone pay to get losses? that is the only part of the Abu-dhabi story that doesnt fit for me.

It is a lot more logical that someone like TSMC or UMC or so could get into a agreement (not actually buy) AMD fabs included the future NY fab and manage them, so they build the fabs and have to renew the machinery to get new production process. But this still has a ton of issues to solve.

This is POTENTIAL LOSS. As AMD fabs are currently very busy, there should be no loss, but the potential exists. This is the key point of ASSET LIGHT...you are light in stocking ASSETS that DEPRECIATE. Intellectual property rights typically do not depreciate in value, but physical things other than buildings do. I am sure that AMD owes Germany somewhat for the loans to build the current fabs, if memory serves me right, so some debt would be gone along with the purchase of the fabs. Money from the deal lines the execs pockets, the remainder goes to pay down debt, and all is well, as they can then lay off yet more staff, and only keep key personal for future development, and likely only a few engineering teams, and some sales and execs.

In the end, we all knew this was going to come, Hector has left now that it's non-reverseable. I wish German governemnt could somehow find HIM personally liable for any loss AMD posts, becasue I think you could find most of that money in his possesion...AMD wouldn't be bleeding so much if thier execs took a severe pay hike...



And Miss Banana...there is LOTS of companies that have gone "Assest Light" before....but maybe only 1% still exist. Another way to explain assest light is selling of company property to pay teh employees...and this is exactly what is happening.

[Bo_Fox]

AMD's 65nm fab still cannot make those 65nm Athlon 64's as good as the 90nm ones were (with 2MB of L2 cache)! UGH, I could never understand that...

[JumpingJack]

AMD's 65nm fab still cannot make those 65nm Athlon 64's as good as the 90nm ones were (with 2MB of L2 cache)! UGH, I could never understand that...

Conventional CMOS scaling hit it's classical limit at the 130 nm to 90 nm node. In order to show any gains at 90 nm, both Intel and IBM/AMD had to implement stress engineered transistors, as it shrinks to 65 nm more of the same is required, unfortunately, the bang for the buck from stress diminishes as you get smaller. AMD threw everything but the kitchen sink into the 65 nm process, 4 different stressing technologies, and it still hasn't produced top bins to match 90 nm ....

On the flip side, Intel's 65 nm process was barely matching the 90 nm top bins as well.... the only comparisons you can make were to compare the Prescott/smithfields to the cedermill/presler ... and if you recall, neither did the 65 nm netburst exceed the 90 nm clocks... both companies got the power benefit, but the Fmax for each was not quite as forthcoming.

It's simply because simple geometric scaling doesn't work any more... the scaling has hit the limits... to get more it takes new materials and radical changes.

Jack

[Rammsteiner]

I hate it when people don't use their brain before posting.

You know that when you talk out of your ass you dont need to use your brain right?

Also it's not good if you hate your self.

Not exactly cheap either.Compared to the US , I'd think they're more expensive by any standard.
Factor in the social benefits and it's a bad choice to produce in Germany in Euros and sell in $.

Germany is globally indeed not the cheapest option to produce. But Germany is by far not expensive. Hell, even NL is already going to be more expensive.

Im not sure but Norwegian/Sweden/Denmark would be more expensive as well. Also, maybe it was cheaper to produce in Germany and ship from there around Europa instead of producing in China and ship it all back.

It's bad to sell in dollars, but producing in dollars might be just as expensive at the very same time.

Let's assume a new entity owns the FABs.Will they produce something else there or just stick to AMD CPUs ? Obviously the former option.

I think the new owner would be more than happy to produce AMD chips since they've a guaranteed customer. Also, I dont think AMD sells their fab to the first person they meet. I think their will be a contract at least binding the fab to produce x years y quantity.

If you produce something else you need a different process because different features are needed.
Can anyone afford to have multiple processes tuned for different markets ? Maybe TSMC and UMC.
Otherwise you have a generic one.

Depends on the capacity you got to produce for said product. If they've multiple processes around they've a widespread income. Just like farmers basicly.

So AMD produces CPUs in house ; those are registered at production cost.
AMD sells the FABs and outsource its CPUs : now you pay production cost + profit margin.

Economics 101.

We were that far Sherlock. Now the question of more importance was, did AMD produce max capacity 24/7 or not?

You can search how much a new FAB costs , add in depreciation , labour deals , etc and you can put a very rough estimate.

Because theory is better than practice? There are so manu uncontrollable variables in there, that's just rubbish if you go that way it is.

Why not wait and see before making such absolute claims.

Well, isnt it true then?

Yes we'll have to wait and see for the result, but the goal is pretty much clear, no?

[Miss Banana]

Savantu your speculation is presented as truth, even though you don't really know what you are talking about.
This new move by AMD is far from ideal, but result of a situation that is far from ideal. What effects it will have remains to be seen.

I just hope AMD will not stop to exist at some point, this would benefit noone.

[savantu]

You know that when you talk out of your ass you dont need to use your brain right?

Also it's not good if you hate your self.

The genius decided to get involved. * applauses*

Germany is globally indeed not the cheapest option to produce. But Germany is by far not expensive. Hell, even NL is already going to be more expensive.

Im not sure but Norwegian/Sweden/Denmark would be more expensive as well.

Why not give the Moon or the North Pole as a counterexample ?
Did it ever crossed your brain that you need an infrastructure and qualified work force for such a large project ? Germany is an engineering power house , that's why it was chosen.

Given dumb examples of more expensive countries which are not on the semiconductor train is an example of not using ones brain.

Also, maybe it was cheaper to produce in Germany and ship from there around Europa instead of producing in China and ship it all back.It's bad to sell in dollars, but producing in dollars might be just as expensive at the very same time.

That's some seriously twisted logic in there.What are the labour costs in Germany vs. China ? Or the utilities cost ?
Did it occur to you that Asia Pacific is the main growth region of the world ?

Why do you think Intel builds a FAB in China ?

I think the new owner would be more than happy to produce AMD chips since they've a guaranteed customer. Also, I dont think AMD sells their fab to the first person they meet. I think their will be a contract at least binding the fab to produce x years y quantity.

No sh*t Sherlock ; who would have thought of that ?

The new owner will charge AMD more ( to recoup its investment and fuel future R&D ) and will fab other products in order to get a lower fixed cost per product.

Depends on the capacity you got to produce for said product. If they've multiple processes around they've a widespread income. Just like farmers basicly.

Probably a new process is worth as much as a few tons of seed.

A new process my dear Watson costs $1-1.5B to develop.To get an ROI on that , you need revenues 4-6x larger.Nobody would develop a specialized new process that will bring a few hundred million $ ; it's not worth it.For such a small foundry it's obvious they will stick to a generic one.

Heck , even TSMC produces Nvidia parts in a generic 65nm process.

We were that far Sherlock. Now the question of more importance was, did AMD produce max capacity 24/7 or not?

Not even close ; this was mentioned clearly in the last earnings conference call.

[Miss Banana]

Dear Savantu,

You said:

germany ... the most labour expensive country in Europe

which is complete and utter bull.

Then Rammsteiner made the far more nuanced and truthfull comment:

Germany is globally indeed not the cheapest option to produce. But Germany is by far not expensive. Hell, even NL is already going to be more expensive.

After which you blast him with this comment

Given dumb examples of more expensive countries which are not on the semiconductor train is an example of not using ones brain.

This is disrespectful and silly of you, since your points have been poor, the supporting sources or explanations for your various "know it all" statements nonexistant, and your attitude mocking.
And putting smilies after sentences in which you feel you have dissed someone really makes you look a lot younger than you really are(unless you really are 12), regardless of whether or not you actually managed to diss someone, which is mostly not the case.

[Shintai]

Dear Savantu,

You said: which is complete and utter bull.

Then Rammsteiner made the far more nuanced and truthfull comment:

After which you blast him with this comment

This is disrespectful and silly of you, since your points have been poor, the supporting sources or explanations for your various "know it all" statements nonexistant, and your attitude mocking.
And putting smilies after sentences in which you feel you have dissed someone really makes you look a lot younger than you really are(unless you really are 12), regardless of whether or not you actually managed to diss someone, which is mostly not the case.

He is right. Try find the needed workers with the skills in those higher payed countries. Remember volume is the key here. We couldnt even build a fab here in Denmark since it would be delayed 2 years or so just due to lacking construction workers. Then you could use 10 years to try and hire some 5000 employees for it.

And Rammsteiner didnt make a nuanced and truthful comment. It was just utter rubbish. Unless this is Xtremefictions.org

Anyway, these changes for AMD means they can most likely fill some of their unused fab capasity with other products. Maybe even pull the Radeon GPUs home. But one thing is sure, this is very very bad for their CPU business future. Unoptimized processes, shifting volume depending on outside contracts. maybe even have to shift some around to foreign foundries aswell.

I wonder what company will keep the debt or how much each. Not even to talk about who will invest. Or if its just open the AMD fabs for outside contracts to utilize their lines for something.

[[XC] riptide]

Less personal BS, and underhand snide comments in these posts boys...

[Rammsteiner]

Why not give the Moon or the North Pole as a counterexample ?
Did it ever crossed your brain that you need an infrastructure and qualified work force for such a large project ? Germany is an engineering power house , that's why it was chosen.

Given dumb examples of more expensive countries which are not on the semiconductor train is an example of not using ones brain.

So NL is like in the middle of nowhere? Hell, NL is a lot better than Germany and yet they chose Germany. And it's not like NL isn't an 'engineering' powerhouse. So obviously Germany had to offer something awesome to chose it over NL.

That's some seriously twisted logic in there.What are the labour costs in Germany vs. China ? Or the utilities cost ?
Did it occur to you that Asia Pacific is the main growth region of the world ?

Why do you think Intel builds a FAB in China ?

Well, do you know? And still, shipping of recources and productions costs a lot.

And whether Intel is building a FAB in China or not, Intel is not like 'the example' for the rest of the world of what the best thing to do is.

And duh, of course do I know Asia is pretty much of high interest nowadays, doesnt change that because of that there's nothing in the west anymore?

No sh*t Sherlock ; who would have thought of that ?

The new owner will charge AMD more ( to recoup its investment and fuel future R&D ) and will fab other products in order to get a lower fixed cost per product.

What's new? You're exactly saying what I said

Probably a new process is worth as much as a few tons of seed.

A new process my dear Watson costs $1-1.5B to develop.To get an ROI on that , you need revenues 4-6x larger.Nobody would develop a specialized new process that will bring a few hundred million $ ; it's not worth it.For such a small foundry it's obvious they will stick to a generic one.

Heck , even TSMC produces Nvidia parts in a generic 65nm process.

Damn, do I really have to chew everything before you can swallow it or what. Of course there's homework to be done to investigate if there's profit to be made out of other products. My dad is some sort of project manager over various plants in Europe. So no need to explain me that.

But yet, it depends on the 'will' and the 'can'. It's risky to invest loads of money in another production line but if you can make profit out of it after 5~10 years it's a matter of willing to do it and it's understandable if you actually wont (it's risky after all).

Not even close ; this was mentioned clearly in the last earnings conference call.

Well, there we go. Dont see the point of discussing this all if that was basicly the most important thing.

And oh yes, btw, Im prolly a bad 'fanboy' cause I dont pay any attention about fabs this and stuff that... *sigh*

[Perp]

I bold out the important part.So no,they won't completely sell the fabs,but more co-manage it and share the cost with other big players(like IBM or TSMC).

I'd like to see IBM acquire AMD straight out.

[Shintai]

I'd like to see IBM acquire AMD straight out.

But IBM and their stockowners dont share your idea

[cadaveca]

But this split may make it easier for nV to aquire AMD....

[donitsi]

So amd turns to dickless fabless :banana::banana::banana::banana::banana: Maybe even goign small like via? :/

[informal]

But this split may make it easier for nV to aquire AMD....

Oh man,tell me you were just joking...

[Hornet331]

Oh man,tell me you were just joking...

well if we go after numbers hes not that wrong, if the split of the fabs, thats most of there inventorie. If they they eliminate so much of there value, im sure the sharepirce will reflect that with a course of 2-3$ practically cut the marketcap in half (from 3,5Bn$ to ~1,75Bn$)

[informal]

Why don't we wait for the actual "sell-off" of the AMD fabs (good luck with that) before we proclaim Nvidia a new owner of AMD.But please,don't let me stop you,it is quite amusing to read all of this .

[cadaveca]

49% control of AMD proper and like 85% of abudhabi holdings might be a way to skirt the X86 license problem. They'd have controlling interest overall, but less than what the Agreement from Intel requires.


Cutthroat, sure, but this is nV and Hector we are talking about...they just need to wait for Intel to release VGA so as to not create a monopoly of the market(By using cpu division, could possibly be legality issues)


This is business, plain and simple, and wages are made on deals, not products, for these guys. Let's be honest about this...the industry be damned when you are close to retirement...

[DoubleZero]

I'd like to see IBM acquire AMD straight out.

No need, maybe just take the partnership to another level.
http://www.tomshardware.com/news/amd...ntel,6175.html

[vengance_01]

I bold out the important part.So no,they won't completely sell the fabs,but more co-manage it and share the cost with other big players(like IBM or TSMC).

I think this is for the best. This allows AMD to put more money into R&D and should reduce costs. Bad news is alot of people will probably get laid off.

[Shintai]

AMD turned from harmless to a competitor with TSMC, IBM, UMC etc. I dont think the partnership will be as strong as before there.