The economics of Bulldozer...

[JFAMD]

Marketing. First on the race track, first in the show room.

So, would you buy an AMD processor if we had the fastest $2000 processor, but at your price level intel was faster?

People don't buy based on the race track, they buy based on what they need.

 

[veri745]

So, would [some portion of the population] buy an AMD processor if we had the fastest $2000 processor, but at your price level intel was faster?

The answer to this question is "Yes".

[Rational/Informed] People don't buy based on the race track, they buy based on what they need.

Fixed

 

[drizek]

Yes.

If AMD proved that it could build a CPU worth $2000 that was compatible with my socket, I'd be more inclined to invest in that platform, with the expectation of getting that $2000 CPU for $150 in a couple of years.

On the other hand, if the biggest, hottest, best binned chip can't justify more than $300, then I wouldn't have a lot of faith in the value of my investment.

 

[ed29a]

Marketing. First on the race track, first in the show room.

You are deluding yourself if you think that the vast majority of the people who buy computers know what the hell is a CPU, let alone 990X. The marketing consumers see is the blue man group dancing on a TV ad with an Intel logo stapled to it. Even then, they probably have no clue what the hell Intel does other than that there is a sticker on their laptop with Intel's logo on it.

 

[exar333]

The thing about Intel is that they expect relatively large profit margins on all the products they sell. AMD is fine with smaller profit margins. Considering their position in the CPU market, they can't afford to be picky about margins. They need to price stuff competitively to get people using their products.

This is something I've thought about, though. And you have to keep in mind that the GPU in SB takes up a fair amount of die space as well. If you get rid of that (which Intel plans to do with the SB-E chips), that frees up room for another two CPU cores that you could add without even increasing die size compared to current chips. It seems like it would be extremely trivial for Intel to sell 32nm products with 8 cores or possibly even more, as long as clock speeds were lowered enough to keep TDP in check. Whether AMD will provide enough competition for Intel to bother releasing SB chips with more cores is another issue, though.

Hahahahahahahahaha

Good luck talking to their investors and getting them on-board to embrace lower margins. Everyone wants the best margins they can get, based on the market, of course.

 

[Arachnotronic]

So, would you buy an AMD processor if we had the fastest $2000 processor, but at your price level intel was faster?

People don't buy based on the race track, they buy based on what they need.

Negative. However, if my price point is $500, AMD is kind of out of the picture because you guys only sell $200 chips at your top end.

Right now, AMD is great for budget -- I just ordered myself an Athlon II X3 as a second rig, but if I were building a serious replacement for the rig in my sig, you can bet your bottom dollar I'd be off buying an i7-2600k or an i7-970 right now. The only thing that'd stop me is if I knew Bulldozer would give me the same performance for less $ or more performance for the same $.

But you guys don't release squat about Bulldozer other than marketing hype, so naturally nobody but raging AMD fanatics are going to wait for it.

Why don't you guys at least give reviewers samples and let them benchmark them? Get us excited? It's hard to cannibalize Phenom sales with previews of a $300 part when your top Phenoms are $200...

I'm sorry, man -- I love AMD and in the past they had the best CPUs, but it's really hard to believe that Bulldozer will be competitive given the lack of benchmarks.

 

[Topweasel]

Hahahahahahahahaha

Good luck talking to their investors and getting them on-board to embrace lower margins. Everyone wants the best margins they can get, based on the market, of course.

That's a stupid way to look at it. All people care about is profits. Margin is great when you get it, but there is more then one way to skin a cat. Selling off Glofo and going fabless means that they can chase total sales instead of ASP and still be successful.

 

[Vesku]

Hear, hear to top end Bulldozer previews. Guess they will be closer to 90 days out than to 60 days and are afraid of killing the discounted Phenom II x6 purchasing that is going on right now.

If Bulldozer does perform well I would expect to see AMD climb back up the price ladder towards the $1000 mark with their top end CPUs. I would point to Sandybridge pricing as confirmation that chip sales over $300-400 must be somewhat cliff like on the sales chart. Intel was comfortable with leaving that space for their main enthusiast line, SB-E socket 2011, where they can make additional profits on the high end chipsets that go with it. If the top tier enthusiast sales were of much importance beyond branding, would you see behavior like that?

 

[podspi]

So, would you buy an AMD processor if we had the fastest $2000 processor, but at your price level intel was faster?

People don't buy based on the race track, they buy based on what they need.

Of course not! But I'm an incredibly informed CPU consumer.

My point was that the 990X wasn't really a product Intel sells to make money, but a product Intel sells for marketing purposes. "Intel has the fastest CPU, so all their CPUs must be the best for the price".


Same reason why car companies want their name on the winning cars.

 

[Topweasel]

Of course not! But I'm an incredibly informed CPU consumer.

My point was that the 990X wasn't really a product Intel sells to make money, but a product Intel sells for marketing purposes. "Intel has the fastest CPU, so all their CPUs must be the best for the price".


Same reason why car companies want their name on the winning cars.

Oh trust me all manufacturers understand the Value of a Halo product, but you can't throw away profitability to chase it. AMD's halo product is going to be a 16 core BD for servers. Chasing the high-end gaming CPU Halo would have hurt their ability to make cost effective cores which has more value in the Server market which has more ASP value.

 

[podspi]

Oh trust me all manufacturers understand the Value of a Halo product, but you can't throw away profitability to chase it. AMD's halo product is going to be a 16 core BD for servers. Chasing the high-end gaming CPU Halo would have hurt their ability to make cost effective cores which has more value in the Server market which has more ASP value.

Business halo parts don't really make much sense to me, since businesses are much more pragmatic about purchases than consumers are.

But yes, it is clear to me that BD was designed as a server-part first and foremost. We'll see whether that was a good decision soon enough.

 

[Topweasel]

Business halo parts don't really make much sense to me, since businesses are much more pragmatic about purchases than consumers are.

But yes, it is clear to me that BD was designed as a server-part first and foremost. We'll see whether that was a good decision soon enough.

My point was that the star of the show will be the 16 core CPU. In a 2p setup that is 32 cores and those (2p servers) makes up most of the server sales. Whether its Virtualization or FPU heavy workloads, its going to sell and it would sell better then a $1k gaming CPU, and they can sell several speed grades for the pragmatic companies. Whereas if they went pure performance with the cores, and went lets say 6 core for 12 core MCM setups, I think it have a negative impact on sales compared to a nearly same cored Intel setup.

 

[Lonbjerg]

Business halo parts don't really make much sense to me, since businesses are much more pragmatic about purchases than consumers are.

But yes, it is clear to me that BD was designed as a server-part first and foremost. We'll see whether that was a good decision soon enough.

"Nehalem" was designed as a server chip too...it just has really good IPC and can be used in a desktop too:
http://www.anandtech.com/show/2594/5

The targeted applications here are very important: Nehalem is designed to fix Intel’s remaining shortcomings in the server space. Our own Johan de Gelas has been talking about Intel not being as competitive in the server market as on the desktop for quite some time now. He even published a very telling article on Nehalem’s server focus before IDF started. While many of Nehalem’s improvements directly impact the desktop market, motivating its design were servers.

With Nehalem Intel hit both markets at the same time...so that "Bulldozer" is a server chip shouldn't be an excuse for poor desktop(read: gaming) performance.

 

[krumme]

"Nehalem" was designed as a server chip too...it just has really good IPC and can be used in a desktop too:
http://www.anandtech.com/show/2594/5


With Nehalem Intel hit both markets at the same time...so that "Bulldozer" is a server chip shouldn't be an excuse for poor desktop(read: gaming) performance.

Well its about earning money, not about winning at 200fps vs. 150fps at 1024*768. The days one cpu could do it all is over, therefore fx. the acceleration of Atom. Get used to it.

 

[Lonbjerg]

Well its about earning money, not about winning at 200fps vs. 150fps at 1024*768. The days one cpu could do it all is over, therefore fx. the acceleration of Atom. Get used to it.

Atom is still geared for MID's...perhaps your PC usage can be done with "ARM level performance"...mine can not.

And this still dosn't alter the facta that Nehalem was a server chip...that also rocked in a desktop.
Now the AMS tune is that "Bulldozer" is a server chip...and not a good desktop chip...an excuse, not a home run.

 

[Topweasel]

"Nehalem" was designed as a server chip too...it just has really good IPC and can be used in a desktop too:
http://www.anandtech.com/show/2594/5


With Nehalem Intel hit both markets at the same time...so that "Bulldozer" is a server chip shouldn't be an excuse for poor desktop(read: gaming) performance.

Not my point. All Main Intel products became server chips when they started the Xeon line. For a little while most of them required their own production line due to Cache variances but just like the Quadro and FireGL, they are for all intents and purposes the same chips.

Now its even more like that.

No one is hinting that BD is a bad processor and can't be an effective performance processor. Just saying that compromises seem to be made, to create a server CPU with higher demand, then chasing .001% of the sales by creating a King of the Hill competing chip.

As they learned from Ati with the 4k series. AMD has decided that creating manageable cost effective chips with 90% the performance is better then to fight diminishing returns targeting that last 10%.

 

[exar333]

That's a stupid way to look at it. All people care about is profits. Margin is great when you get it, but there is more then one way to skin a cat. Selling off Glofo and going fabless means that they can chase total sales instead of ASP and still be successful.

I don't think you understand business very well. AMD has serious supply issues, so their ASP is critical to their overall profibility. If anything, AMD and Intel are in reverse roles. Intel couls still make money with a lower margin, but their investors and history demand solid margins for a product to be released to market. AMD, on the other hand, struggles to produce CPUs in quanitity and needs a solid margin per CPU to make money.

If you think they are happy with a smallish marketshare AND low ASP and profit, you are mad.

 

[podspi]

I don't think you understand business very well. AMD has serious supply issues, so their ASP is critical to their overall profibility. If anything, AMD and Intel are in reverse roles. Intel couls still make money with a lower margin, but their investors and history demand solid margins for a product to be released to market. AMD, on the other hand, struggles to produce CPUs in quanitity and needs a solid margin per CPU to make money.

If you think they are happy with a smallish marketshare AND low ASP and profit, you are mad.

Does anybody know how capacity constrained they actually are right now with 45nm parts, and what their max capacity should be once the 32nm ramp is complete?

They've lost a decent amount of marketshare in the last few years. Unless capacity has also gone down, they could potentially have extra capacity to grow into now.

And no, I'm not talking about TSMC's 40nm 😛

 

[Topweasel]

I don't think you understand business very well. AMD has serious supply issues, so their ASP is critical to their overall profibility. If anything, AMD and Intel are in reverse roles. Intel couls still make money with a lower margin, but their investors and history demand solid margins for a product to be released to market. AMD, on the other hand, struggles to produce CPUs in quanitity and needs a solid margin per CPU to make money.

If you think they are happy with a smallish marketshare AND low ASP and profit, you are mad.

I don't think you understand how AMD has benefited from now being Fabless. Before AMD needed every cent they could get their hands on for a new fab or process. But that's also besides the point, as all AMD needs to do is sell stuff to make a profit. That is what matters Profit. ASP/Margin/Marketshare all of these are just industry phrases for the particular methodology for making a profit.

They can sell 200 million Zacate's for 100 million dollars in profit, or 5 million Server CPU's at 100 million dollar profit. It doesn't matter.

In that sense you are seeing a healthy plan for market penetration. Zacate being created on the cheap at TSMC for the sub $50 CPU market. Llano using old CPU tech and including a vid card but still smaller then a 4 Core Phenom II, filling in the sub $200 CPU lines and making a market that doesn't have competition. BD having an array of 4-6-8 core CPU's to cover the sub $350 (the highest point before sales jump off a cliff) where the largest is smaller then the 6-core Thuban. Then the same tech used for that will be used to sell $300-$1200 CPU's.

Their is opportunity for increase in ASP. But stated earlier it should be less of an issue now that they are Fabless. But most importantly they are chasing sales, marketing and designing equipment to be competitive and of value at almost every price point where sales are optimum. I also don't understand where the supply issues are coming from, they have the largest selection of CPU's that they ever have, have already shipped and will be shipping 3 new and different architectures, with the first Zacate already selling over 3 million chips.

The goal isn't to outproduce Intel, the goal is produce a product at the point where supply meets demand. You increase the demand by giving it value. You give it value by not raping the few customers who would pay outrageous prices by pricing it so high that only they would pay for it. I am not saying that a halo product and higher ASP are out of the question, but you don't come in behind your competition, having been a lower performer for the last 2 years and immediately try to price it to the point that few people can experience it.

Also if you are AMD you have to be smart. You can't outspend Intel in R&D, you can't out process them, better to chase the purchasing market and not inflate ASP on desktops by spending even more in R&D to get into that last 5% of the market.

 

[drizek]

AMD isn't benefiting from being fabless. More accurately, they are benefiting from ATIC dumping huge amounts of money on their fabs. They are investing $8 billion to build a fab in Abu Dhabi, and $5.5 billion just this year to expand existing fabs.

Now consider the fact that GlobalFoundries makes only $3 billion a year in revenue, and you can understand how being fabless helps AMD become "profitable". Global Foundries is operating at a huge loss, at least in the short term.

 

[Topweasel]

AMD isn't benefiting from being fabless. More accurately, they are benefiting from ATIC dumping huge amounts of money on their fabs. They are investing $8 billion to build a fab in Abu Dhabi, and $5.5 billion just this year to expand existing fabs.

Now consider the fact that GlobalFoundries makes only $3 billion a year in revenue, and you can understand how being fabless helps AMD become "profitable". Global Foundries is operating at a huge loss, at least in the short term.

They are still benefiting. Its easier for them to use capacity elsewhere, they don't have to pay the money up front or in the form of expensive loan payments to build capacity. Sure the cost after R&D and Fab production of the silicon itself is more expensive. But now you have a manufacturer that only has to worry about selling the CPU for more then the finished product+R&D costs and not for finished product +R&D+ Fab production. It also allows a company like Global Foundries to attract other clients as capacity increases, lowering the amount AMD is paying in to cover Global Foundries expansion and process costs.

That's all when everything is all said and done. As it is right now you are right AMD is practically getting free hardware from Glofo compared to the money they are investing into production.

 

[podspi]

AMD isn't benefiting from being fabless. More accurately, they are benefiting from ATIC dumping huge amounts of money on their fabs. They are investing $8 billion to build a fab in Abu Dhabi, and $5.5 billion just this year to expand existing fabs.

Now consider the fact that GlobalFoundries makes only $3 billion a year in revenue, and you can understand how being fabless helps AMD become "profitable". Global Foundries is operating at a huge loss, at least in the short term.

This is actually why I am outright flabbergasted that AMD hasn't agreed to fab all their GPUs at GF as well. If GF goes under I'm not sure AMD could find the capacity from somewhere else.