What does AMD's acquisition of Ati mean for Microsoft?

Titanio said:
In a future where CPUs generally were GPUs also - across all markets - that would be akin to asking why Intel wouldn't help IBM make Xenon after their own solution was turned down. You'd be talking about a scenario where 'CPU' and 'GPU' no longer existed discretely, at least in most cases, and where AMD was simply making 'processors'. I'm talking about a very tight level of integration that penetrates across their markets. It'd be like asking Intel if they'd put their FPUs into a CPU from AMD or whatever..it just wouldn't be on.

I'm not sure when we'll get to that point. In the mean time if AMD is making platforms with distinct components that are reasonably tightly coupled but not wholly one chip, and/or continue to make seperate dedicated GPUs, then there remains flexibility.

Although I agree that we might see that form of solution, an all in one CPU, that would have far more severe effects of the overall future strategy of AMD+ATi. I can't really see, atleast 5 years ahead that they will only be having that form of CPU solution, it might be one of many CPUs out there, but I doubt that in such time the GPU market will either seize to exists, due to all in one CPUs, or more or less remove them selfs from the GPU market, as they would more or less have to do if they would no longer provide GPU only solutions...
 
I ain't putting a timeframe on that kind of situation, Platon :) But you are right in saying that if that was the direction we were heading in, the GPU-only market would not disappear overnight. But once such options became available, there's then the question of competitiveness and what a customer would want for their console. If a customer wanted that single processor, which is what I'm talking about, I do not think AMD would collaborate with, say, IBM on that solution. They would say, "hey, we have all the expertise you need here, we'll even do a custom job if you give us enough $$". But therein is the inflexibility, because if IBM looks better for their contribution and ATi/AMD would be better in their contribution, you can't really marry them anymore, and the customer might have to compromise on their dreamteam to get the job done.

I don't even know if the longer term future will trend in that direction, of one processor to rule them all etc. But news like this obviously stirs the debate.

There's a worst case scenario, though, which geo is worrying over in this thread, which hasn't much to do with a future where CPUs and GPUs were one, but where AMD/ATi became so platform-centric or integrated-graphics-centric that they ceded ground at the high end - which I guess could jeprodise the contribution they could make to a potential console partner in the future. I'm not that pessimistic though, I doubt that'll happen. We'll either have our ultimate processor, or we'll continue to have strong discrete parts..I think/hope.
 
Last edited by a moderator:
fearsomepirate said:
Let's suppose that MS just really, really likes what IBM can do for them in the CPU space and considers the big graphics players to be more or less equal in terms of what they can put out. What do you think AMD would rather do:

1. Sell graphics chips to MS for their next IBM-powered console.

2. Sit by and watch nVidia sell graphics chips to MS for their next IBM-powered console.


To make the point, even pointier, imagine if AMD provided the GPU and Intel the CPU in X console.

If the console performed amazingly, Intel would get kudozs for being such a brillant CPU company - potentially hurting AMD in this segment.

They would have to factor in the cost of the damage done as far as consumer preception. Whatever increase in revenue they made that quarter would have to be factored against the long term devaluation of their own brand of CPU against their competitor. It might just not be worth it.

And remeber, Nvidia is no longer your primarly competitor - Intel and IBM are. Will it really hurt their botton line if nvidia has a run away sucess in the discreet GPU sector? They are probably more interested in attacking Intel's share of the integrated graphics market.
 
inefficient said:
Will it really hurt their botton line if nvidia has a run away sucess in the discreet GPU sector?

People buy a lot of video game consoles. Weren't around 150 million sold last generation? That's a pretty big market to just say "no" to. And so far, having chips in video game consoles hasn't translated to OMG WIN in any other space. So suppose everyone's so happy with their IBM console CPU's that there's no way they would give them up. You're looking at the following scenarios:

1. IBM CPU + nvidia GPU. This translates into hundreds of millions of dollars you could be making, but aren't, and increased brand awareness for two of your competitors. It further decreases the likelihood that the console manufacturers in the multibillion dollar business will ever have any interest in your parts.

2. IBM CPU + something no one cares about. This translates into hundreds of millions of dollars you could be making, but aren't, and increased brand awareness for one of your competitors.

3. IBM CPU + ATI GPU. This translates into hundreds of millions of dollars in your bank account, increased brand awareness for one of your competitors, increased brand awareness for one of your products, and leaves the door open to someday seeing an AMD + ATI console.

I don't see how the first scenario helps AMD more than the third.
 
Titanio said:
I'm talking about a very tight level of integration that penetrates across their markets. It'd be like asking Intel if they'd put their FPUs into a CPU from AMD or whatever..it just wouldn't be on.
That's somewhat different though, because both can make FPUs. If insteadof components you think expertise, it's more a case of loadning out engineers with experience to work on a collaborative project. It's like IBM working with Toshiba to produce Cell, though not as wide reching for sure. I still think if MS are making XB9999, and AMD say 'we've got this great integrated chip ready' and MS say 'no thanks, we're using IBM for the CPU aspect. But we really like ATi's graphics expertise and want to do a deal worth $200 million + royalities for some engineers to work with IBM to create a custom solution just for our platform,' ATi are more likely to say yes (depending on other deals and revenue options) than turn down $hundreds of millions because MS chose IBM CPU core over AMD core.

I think the world is moving more away from products and more towards knowledge. We're seeing loads of collaboration these days, with compnaies pulling on the strengths of others to create better platforms than they could create on their own. If you're not able to work in such partnerships, and only offer your own standalone solutions, I think you're going to miss out.
 
You may be right Shifty, but if this was the general direction processing was heading in, AMD's competitors would have to move in that direction too (and it seems Intel could be certainly..). So it wouldn't necessarily be all that different from the FPU analogy. Even if, say CPU company X did not move in that direction and was left with only 'traditional' CPU-related technology to offer, AMD would probably call your bluff and tell you to stop using archaic processors and to come use theirs ;) I don't think they'd be very interested in potentially giving that other company a leg-up in getting to where AMD already is either, even if the chip was to be for one product only (as their competitor would likely gain from the expertise shared by AMD during the project).
 
fearsomepirate said:
People buy a lot of video game consoles. Weren't around 150 million sold last generation? That's a pretty big market to just say "no" to.

That would be 150m consoles sold over 5 years right? And all the consoles combined from Sony MS and Nintendo.

Do you know how many PCs were sold just the first quarter of 2006?
57 Million link

We could be looking at 200mil+ PCs sold per year. And god knows how big the market is for server and distributed comupting chippery. AMD wants a chunk of that from intel.

A deal with MS for a console chip would "optimistically" sell a mere ~10mil units per year. Is it really worth it?
 
Last edited by a moderator:
inefficient said:
That would be 150m consoles sold over 5 years right? And all the consoles combined from Sony MS and Nintendo.

Do you know how many PCs were sold just the first quarter of 2006?
57 Million link

We could be looking at 200mil+ PCs sold per year. And god knows how big the market is for server and distributed comupting chippery. AMD wants a chunk of that from intel.

A deal with MS for a console chip would "optimistically" sell a mere ~10mil units per year. Is it really worth it?


The vast majority of those PCs are not "gaming" PCs but PCs for offices which will only ever run MS Office applications and proprietary business applications. So the comparison is not very useful.
 
inefficient said:
That would be 150m consoles sold over 5 years right? And all the consoles combined from Sony MS and Nintendo.

Do you know how many PCs were sold just the first quarter of 2006?
57 Million link

We could be looking at 200mil+ PCs sold per year. And god knows how big the market is for server and distributed comupting chippery. AMD wants a chunk of that from intel.

A deal with MS for a console chip would "optimistically" sell a mere ~10mil units per year. Is it really worth it?

i would not be surprised if the margins a GPU vendor gets on a console GPU were 5x - 10x the margins they'd get from IGPs that would normally go into the overhelming majority of those (roughly-estimated) 200M PCs a year.
 
london-boy said:
The vast majority of those PCs are not "gaming" PCs but PCs for offices which will only ever run MS Office applications and proprietary business applications. So the comparison is not very useful.

Since when is AMD's target segment gaming PCs only?

It's comparing the pie AMD wants a slice of, compared the pie the console segment is offering as far as revenue.

Even for present day ATI and NVIDIA, selling console chips is just gravy. It is neither the meat or potatos in their main revenue source.
 
darkblu said:
i would not be surprised if the margins a GPU vendor gets on a console GPU were 5x - 10x the margins they'd get from IGPs that would normally go into the overhelming majority of those (roughly-estimated) 200M PCs a year.

We were talking about the potential for AMD sales in CPUs not GPUs. CPUs are by far the larger business.

Alowing the pairing of a Intel CPU to a AMD GPU in a console may be disastorous to their perception in the CPU market. Bad enough to the point where whatever revenue they gain from the deal would simply not be worth it.
 
inefficient said:
We were talking about the potential for AMD sales in CPUs not GPUs. CPUs are by far the larger business.

Alowing the pairing of a Intel CPU to a AMD GPU in a console may be disastorous to their perception in the CPU market. Bad enough to the point where whatever revenue they gain from the deal would simply not be worth it.

ah, sorry. my bad.
 
What exactly is the benefit to ATI of this deal?

I know AMD needed ATI to combat Intel's integrated graphics (which Rein says is responsible for killing PC gaming, he must be pissed about this development!), but ATI gets what out of this? Except for less flexibility?
 
Titanio said:
Even if, say CPU company X did not move in that direction and was left with only 'traditional' CPU-related technology to offer, AMD would probably call your bluff and tell you to stop using archaic processors and to come use theirs ;) I don't think they'd be very interested in potentially giving that other company a leg-up in getting to where AMD already is either, even if the chip was to be for one product only (as their competitor would likely gain from the expertise shared by AMD during the project).
At this point I'd say that's a very unlikely hypothesis. Any major player who's sitting still why the market completely around them is going to get burned, and quite rightly too. If the integrated uniprocessor is the future, they'll all be researching and developing over the coming decades to produce that. I can't imagine anyone's going to end up in 10 years with no development whatsoever in that field when it's what everyone wants. Well, someone might, and then they'll go bust, which is what happens!
 
heh, I was about to make a thread along these lines:
"what does the AMD acquisition of ATI mean for future consoles?"


but since its already been made, i'll kick back with my coffee and read the thread ;)
 
Megadrive1988 said:
heh, I was about to make a thread along these lines:
"what does the AMD acquisition of ATI mean for future consoles?"


but since its already been made, i'll kick back with my coffee and read the thread ;)

I do wonder if AMD will try hard to peddle its CPUs along with (ATI) GPUs in the 2010+ generation consoles... I don't imagine Sony would be up for it (it's possible, I guess, depending on how this generation ends up for them), but the others might be targets for such a thing.
 
Titanio said:
It's an interesting analysis.

I do wonder what happens in 5 or 6 years if you go to AMD and ask them to collaborate with you on a CPU from someone else (e.g. IBM), in terms of integrating some graphics acceleration or providing software/IP or whatever for your one-big-monster-cpu/gpu combo. I've a feeling the answer would be no, followed by an invitation to look at their own solutions. It could really limit the options of some, since you don't have the same flexbility as before. A one cpu/gpu-future means you may have to buy it all together from one place, which may not suit.

I guess it's good for Sony that they aligned with nVidia though. Although who knows, maybe nVidia could be bought up in the interim too.
AMD is not like Sony when it comes to pushing their own tech. ;)

One of the key reasons AMD bought ATI was their partnership with Microsoft, they would be foolish to abandon or even risk it.
 
Back
Top