"Will ATI Be Broken Up?"

Something to bear in mind is that graphics companies and CPU companies don't necessarily have the same analysts following them, or that some investment companies may cover one but not the other. What we have now is ATI being rolled into AMD and its not necessarily the case that all analysts that follow AMD actually understand as much about the graphics market as they do the CPU market. There is going to be a period of adjustment for the analysts and some education to be done as well.

 

From the same arguments it's probably more likely to end up with opposite result. Good brand = more money, and how can you get a good brand? By having good/best HE-card. How I see this is that ATi does what it already does, and in addition to that, ATi+AMD will make those integrated solutions to compete with Intel and Intel's very low-end graphic-solutions.

 

I agree: it wouldn't make sense to spend $5B just to make a 'fusion' CPU+GPU. That said, I'm not convinced they will succeed: tech mergers are extremely difficult to pull off.

There's the period of instability that usually lasts, say, 18 months. During the months after the announcement but before the actual merger, there's a large amount of FUD about future prospects within the company without a lot of possibilities to do something about it: you have 'merger' teams to prepare everything, but they're only allowed to share so much and real decisions cannot be made (or explained.) I've gone through this a couple of times and it can be very demoralizing.

The real fun starts when the merger has closed: at the higher level, execs from different sides start to engage in political chess games to gobble up as much power as they can, making decisions that are not always logical for the engineers (delaying tactics, releasing information to other groups at a slower pace than necessary etc.) Very frustrating all that. Managers are looking for redundancies: they never say there'll be cuts in the engineering staff, but there always are... Architecture staff is usually safe, but you don't need *2* teams desiging chip IO pads or custom memories or PLL's, do you?

There's the differences in design methodology, one side thinking they're doing it better than the other and vice versa. Tool unification is always a way to save money (and most of the time a sensible thing to do), but having sometimes years of specific experience with tool A and being the recognized expert at it, it's hard to abandon all that and switch to something else. Resentment...

And finally (and very important in this case, I think) there's the loss of feeling to work for a special company. A handful of top architects are very important to set the direction and drive the next architecture. They were probably there right from the start of ArtX. They probably made a sh*tload of money when they were acquired by ATI and hailed as the saviors of the company (they were). Joining ATI was the reward of a few years of very hard work and a recognition of their talent.

This acquisition is the opposite: AMD acquired ATI because it was a lot cheaper than Nvidia (from a technical and market point of view, acquiring Nvidia would have made more sense.) ATI dropped the ball, margins dropped, they're just barely making money. This is not the glorious entrance of, well, the savior.

I've been on both side of an acquisition as just an engineer and you can't believe how much difference it makes: you see that the managers that you really respect and trust (yes, they do exist!) don't have the leverage they used to have. And over those 18 months of instability, they are being replaced one by one by someone of the 'other' management.

So suddenly you find yourself working for a 'big' company. Maybe not as big as Intel, but big nonetheless. Your personal contribution will have less impact on the bottom line (even if former ATI gets its margins back to acceptable, the overall impact will be diluted). Even marketing adapts accordingly as super aggressive campains are frowned upon by the laywers. How uncool is all that?

And then one or two top architects decide to leave. Or take a year of absense. Or just become less prolific. And you get a great offer at this sexy new startup. Why stay?

It's not impossible to avoid all that, but it requires a lot of energy that should be spent on getting the next kick-ass GPU out of the door, DAAMIT.

 

It DOES make sense for AMD to spend their kind of money because they want the ATI's chipset tech. They get all of their patents, everything. Now they can dismantal whatever they want and use whatever people they want in whatever fields they want.

I agree with your statement somewhat in that it compliments mine. I do not think they're be able to pull their intergration off as well as they'd like. And the sad part is, if they fail,(and they possibly give up on high end), AMD has much to fall back on. By high-end I mean their complete intergration of high-end graphics with their microprocessors. As I said before, there'll be no more cards like there are now to pop into your computer after 2007.

There have been reports that said AMD had to buy ATi to survive. This may have been true, who knows? But I don't think ATI had to sell out. I'm not 100% sure, but ATI were going along quite nicely. They had(and still do) a huge fan base. The execs got greedy, and or played it very safe, etc. Who knows? All I'm saying is I think it's bad for consumers because now it's very possible ATI's tech in high-end cards will go by the way of the doe doe bird.

ATI's logo is gone. The ATI name site is gone. All replaced with AMD. Not hard to see where this one's going.

If AMD had simply bought the company but left everything in tact for ATI to make it's own cards, etc. And AMD would simply have access to ATI's patents and designs and possibly a few of their technitions, then I would think it not so bad. But that's not what's happening. Watch what happens in 2008, you'll see.

 

http://www.theinquirer.net/default.aspx?article=35352

 

If they only needed the chipset, there are other players that could have been bought for a much lower price. They needed a GPU to integrate into the CPU for low cost notebooks and low cost desktops.

An integrated CPU/GPU will never be the high end, because you'll have to make technical comprimises that you otherwise don't have to make.

I think you'll be wrong about that.

No, but they did have to get their act together wrt margins. You can't run a semiconductor business on 30% margins.

They weren't, by Wall Street standards.

Irrelevant. Those fans are fans because of certain technical features: higher IQ, (slightly) better performance here or there. Advantages that can easily be eliminated from one generation to the next.
And OEM's care even less.

You're thinking like a fan again. What difference does it make to buy an AMD gpu instead of one from ATI?

I don't think AMD will kill off the high end. There is quite a bit of money in there after all and it's needed as a base to build lower end derivatives. My main concern is still the integration process: the name change is usually the least of worries for employees.

 

I don't believe that the AMD/ATI merger will result in no further high-end development of ATI discreet chipsets.

I'm basing this on examining ATI's Torrenza vision in its' entirety. Looking at this, you see AMD building a modular architecture where there are several opportunities to interface "accelerators" with the CPU. 2 of these are on the CPU die and package. These would obviously be lower end.

Further on down the line though we see the HTX slot. This is a Hypertransport-based connection between a discreet "accelerator" card and the CPU (and via the CPU's memory controller to main memory, as well). I guarantee we will see an ATI graphics card designed for the HTX slot. And I think that if AMD hadn't purchased ATI, this would not necessarily have happened. Given the amount of hardware/software development that this will entail and that this will be AMD-only, even if it provided huge performance improvements this would have been a tough sell to an independent graphics chipset manufacturer.

Now, if ATI can provide an extreme high-end part to fill this slot and it creates a competitive advantage for AMD as a gaming platform Nvidia will be forced to follow suit.

AMD/ATI is a better fit than some seem to realize. Aside from the fact that AMD can now fully realize the Torrenza platform without having to rely exclusively on it's partners it also allows them to provide solutions from the HPC sector all the way down to CE devices. And all x86-based.

As an aside it occurs to me that AMD is actually more tied to x86 than Intel is.

 

ATi isn't going anywhere. Even if they phase out the ATi name the same people are still working on building new cards and AMD seems to be encouraging 'em to rather than cutting them back.

 

I said chipset, patents and everything. I didn't say only chipset, that's what you read. Also, you made my point, they only want the lowend graphics, not continue with the high end. This is what I've been saying.

Don't tell me, tell them. This is what they think they can do and I'm saying it won't be successful as they would like it to be. When/If they fail they'll only intergrate for medium level and low end graphics.. my point is still made. This merger is bad for us consumers who want high-end graphic solutions.

Well, here we can agree to disagree. You may find low-end cards still floating around for a little while longer than 2007 only because the profit margins are high on these cards. ..I give you that part...but high-end cards, no, I don't see it.

I don't think they needed to sell out. That's what I'm saying. There's companies that have been around for years with low profit margins. Like I said before, maybe the CEO's got greedy. They were more interested in money than in the welfare of the company and it's products.

They weren't dying. They have many contracts, xbox360, etc. to keep them going.

Not irrelevant. Looks at Nvidia fans. I know people who blindly buy their products no matter what the IQ looks like. And your "slightly better performance here or there" will no longer be viable once ATI's(AMD's) high-end graphics are out of the market.

Like I said, after 2007 you won't find a high-end gpu, only mid to low-end cards.

You just said their(ATI) margins are a low 30%. It's because of their high-end. It's their mid and low-end that make the high margins. You can't say ATI was dying because of low margins and needed to sell out and at the same time say AMD will keep their low margin high-end cards around. So which is it? AMD doesn't want to keep a losing cash cow around, what for?

 

You don't understand : THAT'S THE END OF THE WORLD.
Hum ...

The best answer to that kind of fear is that it's not rational.
SIS and VIA make chipsets and integrated graphics, it would have been cheaper to close a deal with one of them if AMD just wanted to make low-end solutions.
They went for ATI because they wanted a serious player and a foothold on the discrete market
I don't really see why they would S3 it now.

That's a tribute to ATI's marketing to see how people loose their senses over a logo color change.

PS : the ATI smiley is still red here.

 

So now ATi is loosing money on high-end ?
That's new.

And even so, that would be no reason to step out.
If you are a car-maker you might run in formula one even if it looses a few bucks.
It's still worth it because you gain some brand awareness, you develop some techs that you will be able to use in your mainstreams products and it's good for the morale (always handy when people are your assets).

 

Less than 30% margin = going downhill according to silent_guy. I already said, it wouldn't/shouldn't cause them to desparately sell out. ATI was doing low profit high-end ..but that's not what AMD have planned for the future, imo.

 

IMO, AMD paid a lot of money for ATi, and as such they will naturally want to maximize the value they get for their money. Saying that, it seems to me that ATi's value would be considerably lessened without the reputation that proceeds from high-end competitiveness.

ERK

EDIT: somebody snip this last page to the thread in the Industry forum, please.

 

And what's your source on the supposed "fact" as per which the Radeon X1900XT and above have profit margins below 30%? No, please, I'm listening...


Uttar
P.S.: Your reasoning is even more ironic taking into consideration the fact that ATI's weakest margins are in their IGP business. I guess AMD is going to get rid of all Motherboard-GPU and CPU-GPU integration plans now because of that, right?

 

My fear is that the first thing to be compromised in case amd needs money, is the high end gpus. Its that simple. What amd wants is to become intel. And in order to do that, they need a complete platform. Sure, it will be nice if they can keep producing high end gpus, but if things start going wrong, high end gpus and high end r&d will be the first thing to be sacrificed in order to save tons of money.

Ati couldnt sacrifice high end gpus cause they were the ones who were getting ati all the street credit/advertisement/hype/whatever. Amd can do that though with minimum casualties.

And amd does need money atm. They need to greatly increased their production capabilities and production technologies in order to catch up with intel. Amd knows that the meat is on the low end integrated gpu and complete pc platforms and they will do anything they can in order to conquer that market.

This merge has nothing good to offer to ati(except for the money to its shareholders). At least if ati merged with intel, they could use intel's manufacturing power and get propelled ahead of nvidia. But amd wont be manufacturing ati gpus, not anytime soon at least(they can barely keep up with their own production).

 

Not sure where I said X1900XT and above have profit margins below 30%. Please find that quote from me. I never mentioned the X1900XT in anything. It was silent_guy who said "You can't run a semiconductor business on 30% margins.", not me. Read the thread again.

WHAT? I said that's ALL they're going to do in the future. I think you have my quotes confused with someone else's.

 

Please... Stop ridiculizing yourself. You did not mention X1900XT directly, you said "high-end". If you have another definition of high-end that is SIGNIFICANTLY different, please let me know, as well as the rest of the universe.
You also clearly said that ATI's high-end margins were low, and considering you quoted silent_guy, you thus implied that you thought they were below 30%. Stop defending yourself by taking things literally when what I'm saying is, imo, relatively clear. But feel free to disagree with that the fact I'm clear enough for you, I guess.

You, sir, need to lookup the word "sarcasm" in your dictionary. I was clearly implying that you made no sense, because you said they'd quit the high-end market due to low margins, yet their low margins current are in the IGP market, which you suggest they're going to focus entirely on! You're contradicting yourself massively there.


Uttar

 

True, I agree with NIB on this one. The first thing to cut loose is ATI's high-end graphics if things go bad.
AND and second bold highlight, Also true, there's nothing good for ATI in this buyout except for the CEO's and shareholders.

It could[/] have been good for ATI if someone else had bought them as a subsidiary company, say, Intel. (at least it would have been far better for them than an AMD buyout imo)

 

A slight tangent on the discussion

While I don't have any real idea whether or not ATI will be broken up, I did have a couple of questions that might help the discussion.

1) Intel has a graphics core. Why don't they focus on high end graphics? Is it a lack of resources (cash or personnel) or financial? Do they simply feel it's not a core competence and therefore irrelevant?

2) Given the multi-core phenomenon, are there skills from GPU design that AMD is looking for aside from the graphics processor itself, such as how to design highly parallel systems, etc.?

3) Is there a possibility for AMD to spin off ATI as a sub-unit and use this new group's technology for its upcoming products? So rather than having to license graphics technology, own it and have it specially designed for simple integration. Shared learning and technology while keeping separate business and financial focuses. Would it make sense for this new unit to continue building high end discrete solutions?

4) Will the market for high end discrete solutions change, and if so, will the new environment promote a monopoly, duopoly, oligopoly, or eventually a pure competitive market?

Unfortunately, I have no answers.

 

NO, no, you misread what I wrote. I said "Less than 30% margin = going downhill according to silent_guy." Note the bold. I didn't say it was from me!!! He's the one that thinks ATI's margins are less than 30% and thus was not doing very well in the semiconductor business. ..and thus was a good reason to sell to AMD. Go read his quotes. I don't care if ATI's margin were 5% or 2000%. I was saying I don't think ATI should have sold out and that they could still operate at low profits.(even IF 30% is low for some people)

I said ATI was doing what ^M^ was suggesting, the car company example he gave. They were doing low profit high-end, not low margin. As profit I meant their total high-end profit, not the markup of each individual board. I think you're getting excited over nothing.