Would Have AMD Survived without Having Taken Over ATI Technologies?

[MODEL3]

http://www.xbitlabs.com/news/c..._ATI_Technologies.html

?Both AMD and ATI suffered from inadequate execution in 2007, and it showed in their results. Both got on track by the end of that year, and have delivered on their roadmaps,? said Nathan Brookwood, the principal analyst with Insight 64"

?The ?old? AMD, with integrated manufacturing, but without graphics, would have been challenged to survive last year's global downturn and Intel's new Calpella/King?s Creek/Piketon platforms. The 'new? AMD can survive on lower revenues, and can credibly argue that its integrated graphics platforms deliver a better visual experience than Intel's, even if the Core i3 and i5 CPUs outperform AMD's processors. It would never have been able to make such a claim without the ATI acquisition under its belt,? said Mr. Brookwood

?As you know well, much has happened both in the industry and in the global economy in the three years since the ATI acquisition occurred. Having said that, we remain convinced that the rationale underpinning the ATI acquisition remains sound. Only two companies in the world deliver x86 processors in volume, and only two companies deliver leading edge 3D graphics. AMD is the one and only company in the world can do both - and that's a unique, direct benefit from our ATI acquisition,? said Michael Silverman, a spokesperson for AMD

Analyst Nathan Brookwood believes that the take over actually helped AMD to survive: the company gained new revenue streams with discrete graphics processing units as well as its core-logic sets, which not only bring additional earnings, but also gave AMD technological advantage in terms of graphics processing over Intel: it is indisputable fact that ATI Radeon graphics cores are a lot better compared to Intel Graphics and Media Accelerator (GMA) cores

?I am as disappointed as anybody by the continued push-out of AMD's Fusion CPUs, but can't argue with their rationale for moving them out. Although Intel has multi-chip module integrated graphics now, and single chip integrated graphics next year with Sandy Bridge, both products use Intel's archaic GMA technology that lacks OpenCL and DirectCompute support. This means AMD's products offer better graphics features and greater functionality now, and will beat Intel to market with CPUs that offer OpenCL and Direct Compute, since Larrabee-type graphics don't get integrated onto Intel CPUs until 2013,? said the analyst

 

[Idontcare]

Survived till now, or survived to 2015 or so after a few years of fighting whatever Intel has in store for the markets post Clarkdale?

It was a necessity IMO.

AMD owes their existence to well played acquisitions in the past, K6 and nexgen was pivotal.

If AMD survives the next 10yrs while continuing to exist in the x86 markets it will have been undoubtedly because of the timeliness of their ATI acquisition.

(meaning if they don't substantially divest into becoming primarily a GPU seller or some other market altogether in the meantime and abandon x86 in order to survive by 2019)

 

[Forumpanda]

Talking with no insight what so ever, I felt that AMDs great plan (tm) with ATi was to basically take over the graphics market by dumping money, IP and manufacturing into ATI and hoping the result would be something that could take over nVidias market.

So I think part of the plan died when they had to spin off GF and essentially giving nVidia a chance to compete on an even footing (good for us consumers if AMD/ATI doesn't collapse).

However I think much is still to be said about the acquisition, mainly that turn around times in the GPU worlds are measured in years and that while I am sure the ATI folks are getting some help from AMDs own experts in the field, they are essentially still just refining an old design.
I expect AMD to launch a (relatively) new GPU design when they switch to GF (possibly a shrink of HD5xxx first - as a backup plan), and then we will see where it puts them in relation to what nVidia will have to show by that time.

Given the performance current series of cards coupled with the very (historically) low prices, they are getting what market share they can and at least setting them self up for a home run if they hit something great with AMD expertise + GF manufacturing.

Of course I am just wildly speculating because it is fun, and have absolutely no idea of what is actually going on

 

[Zensal]

I think if Bulldozer turns out to be a flop compared to Intel's Sandy Bridge, then they are done as far as x86 goes. Intel is already killing them in the mobile market and they are moving into the SoC market.

Again, if Bulldozer fails, AMD will probably turn into a GPU only business.

GF will probably go on to succeed with the substantial head start in technology that they have. Not to mention the investors with deep pockets.

 

[sandorski]

It was a good acquisition. When the Economy improves and if ATI can keep competitive with Nvidia, I think AMD will start turning a nice Profit. This is especially true right now as AMD is reasonably competitive in the Processor Market and could possibly turn a Profit in that Market alone, but add in ATI and their position becomes much stronger. I think it should be noted that AMD's CPU Marketshare is approx(last I saw anyway)50% greater than it was in the K6/SS7 days, so despite their lack of Profits they've been in worse situations and survived. The flipside to that though is that Processor Margins have dropped a lot from those days, so that's a factor against them. However, Video Card Margins seem to remain fairly stable and similar to what they have been from those days, with a period of High points and Low points during that period, which doesn't really affect AMD much since they're a recent player in that Market. That said, now that they are a player in that Market, it helps to offset Low CPU Margins, which is a luxury Intel doesn't currently have, although Intels Processor Margins are higher.

Basically speaking, without ATI, AMD would be in a much more precarious position.

 

[CurseTheSky]

From a "what if" standpoint, it's hard to say. If AMD had dumped three quarters of all of that money into R&D and manufacturing costs, produced a superb product, and actually followed through with it (dumping the other quarter into advertising), who knows? Maybe they could have taken significant market share from Intel. For example, the Core architecture is excellent - no one can doubt that. However, as Core 2 was first hitting shelves, if AMD could have churned out something similar to what Phenom II or Atom is now, it would be a whole different ball game. Beating Intel to the quad core market would have gained a significant amount of market share in the server department, and a decent revenue from enthusiasts. Developing a tiny, low power, low cost chip like Atom (Geode on steriods, I guess?) and marketing it the way Intel has could have given them their own little corner of the market to play with for a while. Who knew that netbooks would be so popular?

In the long run, I think the acquisition was a good thing. It diversifies what they're offering, and who they're competing with. Now if Intel churns out a force to be reckoned with, they can cut the costs of their products and focus on the GPU market. As the article stated, they also have the ability to offer platforms that are vastly superior to Intel's offerings - too bad the average Joe doesn't know that. Since that's the one area they can easily claim to dominate Intel, they should at least provide charts for manufacturers and retailers to include with the products, boasting the graphical superiority of their platforms over Intel's.

 

[Accord99]

It seems obvious now in hindsight that it was by far the single worst decision AMD has ever made, massively overpaying for ATI and paying for it in cash. Doing so essentially destroyed AMD's financial health just as Core 2 began to appear and ultimately forced it into the desperate asset-lite strategy it employs today.

 

[Idontcare]

Originally posted by: CurseTheSky
From a "what if" standpoint, it's hard to say. If AMD had dumped three quarters of all of that money into R&D and manufacturing costs, produced a superb product, and actually followed through with it (dumping the other quarter into advertising), who knows? Maybe they could have taken significant market share from Intel. For example, the Core architecture is excellent - no one can doubt that. However, as Core 2 was first hitting shelves, if AMD could have churned out something similar to what Phenom II or Atom is now, it would be a whole different ball game. Beating Intel to the quad core market would have gained a significant amount of market share in the server department, and a decent revenue from enthusiasts. Developing a tiny, low power, low cost chip like Atom (Geode on steriods, I guess?) and marketing it the way Intel has could have given them their own little corner of the market to play with for a while. Who knew that netbooks would be so popular?

FWIW they wouldn't have been able to borrow the money as they did for ATU acquisition just to pump it into R&D. Like most any loan it must be collateralized somehow, and you can't really collateralize future R&D products.

Also the timing means, even in a hypothetical sense, that the R&D products which would have came from investing the money into R&D versus the acquisition IS Bulldozer and 32nm.

These products (both IC design and process tech) take 4yrs, so at the time of the acquisition there was very little to be changed in regards to 65nm/45nm timeline (and device metrics) as well as Phenom and PhII timeline (and specs). Bulldozer is the first architecture to be developed in the environment that could have been different had AMD had cash reserves from 2006 instead of buying ATI.

So a real question here is just how much did the ATI acquisition, compounded with the gross margin erosion in their CPU ASPs brought on by Core2Duo, serve to undermine the scale and scope of Bulldozer's architecture as well as its release timeline. (both BD development and 32nm process development)

What concerns me is that this produces an outcome that says the glass is less than full when it comes to BD, at best we would have had a full glass of BD coming our way, but now we are most assured we are going to get something less than a full glass (or it will just be served well past breakfast but still be a full glass).

Originally posted by: Accord99
It seems obvious now in hindsight that it was by far the single worst decision AMD has ever made, massively overpaying for ATI and paying for it in cash. Doing so essentially destroyed AMD's financial health just as Core 2 began to appear and ultimately forced it into the desperate asset-lite strategy it employs today.

Good point, I had not considered that the asset-lite was a direct result of the knock-on effects of the ATI acquisition...but it stands to reason if ATI purchase sucked up cash and then subsequently things that needed cash (bulldozer development, 32nm development, fab management, salaries, etc) became cash starved and suffered from it then it is hard to argue they did not suffer from the acquisition.

FWIW I consider the asset-lite model to be a positive thing for AMD and for GF. They needed to do something to unlock the value of their process tech and fab production technology.

If they were too cash poor to diversify into more markets that would have then enabled them to leverage that inhouse production expertise then it only makes sense to spin off the fabs. That's actually what you do when you make tough decisions regarding doing what is best for your shareholders in the long-run versus just doing whats best for your employees in the short-term.

(I say this after experiencing my employer, Texas Instruments, having gone thru similar asset-lite(r) strategy of shifting all 45nm and beyond process technology development and production to the foundries...a tough decision that I know wasn't made lightly, and it impacted me negatively at the personal level, but was the best thing for shareholders in the long run)

So I see the asset-lite deal (for a lot more reasons than the scant few I discussed here) to be great for absolutely everyone involved TBH. If it took the ATI acquisition to put AMD decision makers in a cash-poor bind to force their hands to make such a decision then I would actually argue the ATI acquisition proved its worth right then and there, albeit the dividends won't be paid back for another few years.

 

[imported_Lothar]

Originally posted by: Accord99
It seems obvious now in hindsight that it was by far the single worst decision AMD has ever made, massively overpaying for ATI and paying for it in cash. Doing so essentially destroyed AMD's financial health just as Core 2 began to appear and ultimately forced it into the desperate asset-lite strategy it employs today.

You pretty much hit the nail on the head.
If they had bought ATI as an all stock transaction rather than using their "cash on hand" and high interest loans, they would have been in a MUCH better position.

 

[heyheybooboo]

Originally posted by: Lothar

Originally posted by: Accord99
It seems obvious now in hindsight that it was by far the single worst decision AMD has ever made, massively overpaying for ATI and paying for it in cash. Doing so essentially destroyed AMD's financial health just as Core 2 began to appear and ultimately forced it into the desperate asset-lite strategy it employs today.

You pretty much hit the nail on the head.
If they had bought ATI as an all stock transaction rather than using their "cash on hand" and high interest loans, they would have been in a MUCH better position.

Call really agree with any of this.

AMD didn't expend any cash for ATI.

AMD today has more cash on hand than anytime in their history but they are taking on more debt.

Revenues are down but AMD will make $5.5 billion this year.

Current liabilities are down 30% from 2006.

AMD R&D expense is as high as it has ever been.

Nobody likes their level of debt but as I understand it they have until the middle of 2012 to pay it down and refinance. I'm not sure how much of their debt will be assumed by GF.

I'm not playing Rosy Scenario, here. AMD is in a world of hurt but they are not going anywhere. This stuff seems to be a continuation of the same ol' gloom and doom we've seen over the past 3 years.

Rumors persist about the problems not only in the nVidia GPU transition to 40nm but with the future prospects of their chipsets. If their license is not renewed with Apple Jobs will most likely revert back to Intel for desktop chipsets.

And that opens the door for AMD graphic chipsets for Apple notebooks. Let's face it, the nVidia relationship with Apple has not gone well. We live in litigious times but can Intel, nVidia and Apple find anymore convoluted ways to tussle legally?

If AMD advances its integrated graphics chipset line to higher levels of performance with the new sb8xx the door may be opening for them to snag substantial OEM business, put a dent in Intel's desktop marketshare and further erode nvidia chipset expansion.

If this becomes the case the AMD acquisition of ATI is all win, and if not, it's still exceptionally important to their own current and future platform development.

 

[drizek]

It was a great move.

Ignoring the financial side, just the integration between the two is hugely significant. By far the most compelling aspect of AMD platforms, to me, is the chipsets. I used my integrated graphics for 6 months or so before buying a 4830, and I will use it again when I repurpose this system for multimedia or to give to someone else. Intel had the G33 out, which was crap performance wise, and had no DVI. My $80 AMD board had DVI, HDMI and Optical out. Combined with an AMD processor that has better performance/dollar than Intel, it was an easy choice. Moreover, idle power usage with 780G/4850e that was my original setup was extremely low.

 

[imported_Lothar]

Originally posted by: heyheybooboo

Originally posted by: Lothar

Originally posted by: Accord99
It seems obvious now in hindsight that it was by far the single worst decision AMD has ever made, massively overpaying for ATI and paying for it in cash. Doing so essentially destroyed AMD's financial health just as Core 2 began to appear and ultimately forced it into the desperate asset-lite strategy it employs today.

You pretty much hit the nail on the head.
If they had bought ATI as an all stock transaction rather than using their "cash on hand" and high interest loans, they would have been in a MUCH better position.

Call really agree with any of this.

AMD didn't expend any cash for ATI.

AMD today has more cash on hand than anytime in their history but they are taking on more debt.

Revenues are down but AMD will make $5.5 billion this year.
Current liabilities are down 30% from 2006.

http://news.cnet.com/AMD-folds...00-1006_3-6129445.html
Where do you think the $4.3 billion in cash came from?
Do you think the money appeared from thin air?

The only people who've won so far are people like Dave Orton(and other ATI execs/shareholders) who cashed their stock options and bolted the firm due to AMD handing them $4.3 billion in cash instead of doing an all-stock transaction.

If you're a stock holder who considers himself "part owner" in the business, you not only should concern yourself with "current liabilities", you should also be looking at long term debt.
If you're simply a day trader who doesn't hold stocks for more than a few months, then long term debt is nothing you should concern yourself with.

From Google Finance and AMD's filing reports:

Total Cash and Short Term Investments
As of 2006-12-31: $1.541 Billion
As of 2007-12-29: $1.889 Billion
As of 2008-12-27: $1.096 Billion
As of 2009-06-27: $2.514 Billion

Total Debt
As of 2006-12-31: $3.797 Billion
As of 2007-12-29: $5.269 Billion
As of 2008-12-27: $5.074 Billion
As of 2009-06-27: $5.621 Billion

I'm not sure how that's a good thing.

Next time you receive your credit card bill, check your statement.
I borrowed $603.82 from my credit card company this month. (Cash on hand).
My "current" liabilities would only be the $20/month minimum payment I'm required to make.
My total debt would be $713.09(principal+interest) and will take 3 years to pay loan off.

I decide to do a balance transfer to another credit card which paid me $603.82 to pay off my other loan and charged $50 in processing fees for a total loan of $653.82 with the same APR, but only requires a minimum payment of $14.

My new long term debt will be $863.81 ($653.82 principal+$209.99 interest) and will take 5 years to pay off loan.

My cash on hand($713.09) is now more than I've ever had in my entire history($603.82), so what?
My "current" liabilities($14/month minimum payment) is 30% less than it was a few months ago(the previous $20/month minimum payment).
So now I have $713.09 cash on hand, but $863.81 in debt? How exactly is that a good thing again?

Do you see it as a good thing because I have more cash on hand, and my "current" liabilities are down 30%?
I didn't "expend" any cash to pay for my original credit card bill either in this scenario, right.

 

[Accord99]

Originally posted by: heyheybooboo
Call really agree with any of this.

AMD didn't expend any cash for ATI.

AMD spent $4.2 billion in cash, some from a loan and some from existing cash and short-term assets in the purchase of ATI.

AMD today has more cash on hand than anytime in their history but they are taking on more debt.

Sounds like the US Treasury.

Revenues are down but AMD will make $5.5 billion this year.

Revenues aren't profit. AMD will lose hundreds of millions of dollars this year yet again.

I'm not playing Rosy Scenario, here. AMD is in a world of hurt but they are not going anywhere. This stuff seems to be a continuation of the same ol' gloom and doom we've seen over the past 3 years.

It's just that if AMD did not purchase ATI, it would be in a less world of hurt. Or at least, paid for it in stock, or waited for a better purchase price.

Rumors persist about the problems not only in the nVidia GPU transition to 40nm but with the future prospects of their chipsets. If their license is not renewed with Apple Jobs will most likely revert back to Intel for desktop chipsets.

And that opens the door for AMD graphic chipsets for Apple notebooks. Let's face it, the nVidia relationship with Apple has not gone well. We live in litigious times but can Intel, nVidia and Apple find anymore convoluted ways to tussle legally?

The graphics market simply isn't that profitable enough to make a meaningful difference to the health of AMD. The graphics division didn't even make money for the first two quarters of the year.

 

[imported_Lothar]

Originally posted by: heyheybooboo
I'm not playing Rosy Scenario, here. AMD is in a world of hurt but they are not going anywhere. This stuff seems to be a continuation of the same ol' gloom and doom we've seen over the past 3 years.

They could have reduced that hurt by paying for it in an all-stock transaction instead of paying $4.3 billion in cash.
Rather than simply putting band-aid on a wound, it seems all they did was add salt on the wound then put the band-aid on it.

The number of ATI executives who were fleeing the firm after AMD overpaid the company in cash was in itself ridiculous.

 

[Idontcare]

edit: post's original content deleted as relevance is now immaterial. (see Lothar's post below)

 

[imported_Lothar]

Originally posted by: Idontcare

Originally posted by: Lothar
As of 2009-06-27: $9.148 Billion

Wha!? :shocked: AMD's debt is now $9B? Am I not understanding the context that this is taken from or has their debt actually gotten this high?

I haven't followed AMD financials during the whole GF transition as I expected things would get a little wonky on the balance sheets as all that paper debt and paper equity swished around from one bucket in accounting to another...but did AMD really exit the GF deal with $9B of debt to service?

What is that costing them, interest payments wise, per annum?

Oops, it's actually $5.621 billion. Looked at the wrong line.
Post updated and corrected.

 

[imported_Lothar]

AMD completed it's acquisition of ATI on October 3, 2006.
In my post above, all those yearly figures (including 2006) was after the acquisition.

Oh and for comparison, AMD only had a total debt of $1.370 billion on 2005-12-25 (before acquisition).

From $1.370 billion to $5.621 billion debt in less than 4 years?
What was Hector Ruiz thinking?

 

[themusgrat]

Definitely the right thing to do, they also definitely have alot of catching up to do, but the video card section of AMD isn't the thing holding it back.

 

[drizek]

What was Hector Ruiz thinking?

That the profits from ATI would exceed the interest payments on the debt, and that the value of ATI as a company would not go down. This alone is probably not reason enough to buy out ATI, but then there is the matter of Fusion, as well as the ability to custom tailor chipsets and CPUs.

There was also the possibility of fabbing ATI parts in house, helping to prop up the company as a whole(now GF). It still seems pretty likely that the 7000 series cards will be done on GFs 28nm. There is some good money to be made there, particularly if GF one ups TSMCs equivalent. At that point Nvidia will have to either manufacture their GPUs at their competitors fab(possibly as a second class citizen) or put up with using an inferior process. Meanwhile, TSMC loses a huge bulk of revenue, making RnD harder, potentially making it fall further behind.

 

[Idontcare]

Originally posted by: drizek

What was Hector Ruiz thinking?

That the profits from ATI would exceed the interest payments on the debt, and that the value of ATI as a company would not go down. This alone is probably not reason enough to buy out ATI, but then there is the matter of Fusion, as well as the ability to custom tailor chipsets and CPUs.

There was also the possibility of fabbing ATI parts in house, helping to prop up the company as a whole(now GF). It still seems pretty likely that the 7000 series cards will be done on GFs 28nm. There is some good money to be made there, particularly if GF one ups TSMCs equivalent. At that point Nvidia will have to either manufacture their GPUs at their competitors fab(possibly as a second class citizen) or put up with using an inferior process. Meanwhile, TSMC loses a huge bulk of revenue, making RnD harder, potentially making it fall further behind.

+1

We can safely assume the acquisition was only pursued for one of two reasons - (1) either to expedite diversification of the markets that AMD's products served, or (2) enable better defense of existing marketspace that AMD occupied at that time.

I view Fusion as a defensive product, for example, designed to fend off encroaching products from then ATI or Nvidia in the GPGPU space (same rationale I assume motivated Intel to pursue Larrabee) so that ASPs on x86 marketspace products would not erode any faster than usual.

The discrete GPU market further enables AMD to have yet another market to field their chips into. I have no doubt at the time of the acquisition the internal line of thinking was that ATI's products would be produced inhouse in their fabs and that would lower the cost of manufacturing for both the GPU's and CPU's. (again same manufacturing cost analysis behind Intel's motivation to pursue Larrabee)

Not too mention the IGP and chipset boost the acquisition gave AMD. Also we can't, I can't, over-state the value of having unfettered and immediate access to ATI's IP. Lord only knows just how hamstrung Intel's Larrabee really is solely due to Intel having to navigate a well-occupied IP-space given ATI's and Nvidia's existence for decade+ prior.

So at best we can argue the acquisition will serve both reasons, at worst it will have served none, but odds are (2x2 matrix) that it will have at least served one or the other and that is better than the alternative. I like those odds (1/4 maximum benefit, 2/4 one-half benefit, 1/4 no benefit).

The only thing to monday morning quarterback here IMO is: Could AMD have better timed the acquisition itself to have scored a lower price on ATI that better reflected the short-term value of ATI on the books?

AMD's market timing...so much like my own when it comes to playing the stock market :disgust: :|

But what we don't know is just how much of the ATI acquisition was really rushed after the initial failed efforts to buy Nvidia. Once that happened, the failed effort, I'm sure there was a matter of a sense of urgency on AMD's behalf to bag ATI before market speculation transpired and ATI's price got REALLY pumped up.

In any moment of crisis, or situation commanding a sense of urgency, less than perfect efficiency is assured to occur, which is why managers are taught to endeavor to avoid crisis situations thru expending more efforts towards adequate planning and risk mitigation.

 

[IlllI]

i can't say yes or no, but i will say that if AMD goes down then prepare to pay exorbitant prices for intel processors again. with no competition they can basically price them whatever they feel like, and people will have no option but to pay that amount

 

[OCGuy]

ATi is losing money, and AMD is in more debt. Very bad idea to make that deal.

We may be talking about who is going to take over AMD if this keeps up.

 

[Fox5]

Originally posted by: Zensal
I think if Bulldozer turns out to be a flop compared to Intel's Sandy Bridge, then they are done as far as x86 goes. Intel is already killing them in the mobile market and they are moving into the SoC market.

Again, if Bulldozer fails, AMD will probably turn into a GPU only business.

GF will probably go on to succeed with the substantial head start in technology that they have. Not to mention the investors with deep pockets.

90% of AMD's business is cpus, no way they could abandon that. They'll either continue as a low cost provider, or find a niche.
Their best option would probably be mobile. They have a major head start in graphics (and presumably directx compute), and the thermal limits of mobile give them a better chance to match performance/watt.

And GF still needs high performance customers in volume to maintain their process lead. For the near future, that's AMD, and maybe ATI and nvidia to a lesser extent.

I don't think the ATI acquisition can be considered anything but a mistake though. AMD bought ati when it was at its strongest, only to shortly watch it go to its weakest. That money could have been pumped into getting their 45nm process out on time, and maybe a Phenom 2 to compete with penryn.
That, or maybe AMD wouldn't have neglected their mobile market.

 

[Zstream]

Originally posted by: Fox5

Originally posted by: Zensal
I think if Bulldozer turns out to be a flop compared to Intel's Sandy Bridge, then they are done as far as x86 goes. Intel is already killing them in the mobile market and they are moving into the SoC market.

Again, if Bulldozer fails, AMD will probably turn into a GPU only business.

GF will probably go on to succeed with the substantial head start in technology that they have. Not to mention the investors with deep pockets.

And GF still needs high performance customers in volume to maintain their process lead. For the near future, that's AMD, and maybe ATI and nvidia to a lesser extent.

This is where I disagree, AMD & GF only need to make enough money to stay even. When the time is right, we will start seeing a better saturation into the enterprise environment.

 

[drizek]

Originally posted by: Fox5

Originally posted by: Zensal
I think if Bulldozer turns out to be a flop compared to Intel's Sandy Bridge, then they are done as far as x86 goes. Intel is already killing them in the mobile market and they are moving into the SoC market.

Again, if Bulldozer fails, AMD will probably turn into a GPU only business.

GF will probably go on to succeed with the substantial head start in technology that they have. Not to mention the investors with deep pockets.

90% of AMD's business is cpus, no way they could abandon that. They'll either continue as a low cost provider, or find a niche.
Their best option would probably be mobile. They have a major head start in graphics (and presumably directx compute), and the thermal limits of mobile give them a better chance to match performance/watt.

And GF still needs high performance customers in volume to maintain their process lead. For the near future, that's AMD, and maybe ATI and nvidia to a lesser extent.

I don't think the ATI acquisition can be considered anything but a mistake though. AMD bought ati when it was at its strongest, only to shortly watch it go to its weakest. That money could have been pumped into getting their 45nm process out on time, and maybe a Phenom 2 to compete with penryn.
That, or maybe AMD wouldn't have neglected their mobile market.

It is true that ATI was headed for a fall when AMD purchased it. However, they are back in the game in a big way now. They are shipping 40nm DX11 parts and Nvidia isn't. Thats going to be a big boost for both profit margins and market share.