Advanced Micro Devices
How will AMD Survive?

Related Links
Discussion Boards

By Roleplayer
April 30, 2008

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

AMD is not going to survive if it continues executing like it has the past two years, especially on the CPU side of things.

That said, I believe that most of 2007's problems came from two sources. Source one was AMD signing Dell, which caused every spare resource, and some that weren't spare, switching from Fab36 and 65 nm ramping to making enough CPUs to satisfy Dell late in 2006. We will never know how much AMD's ramps were hurt in 2007 by this strategic shift in 2006, but I'm sure that it was more than none. AMD certainly made more money in 2006 because they were selling to Dell, but they seem to have not realized the size of the coming floods and set that money aside for bad times. Thus, layoffs today. It was definitely one of the more clearcut short term profit versus long term goals decisions I can recall, and if AMD does go belly up my first blame will go to that decision.

Source two has to come with expectations. When making a new CPU, it can ship really early in the mask process, which means that it works great right out of the box. It can ship about as expected, 1-2 revisions needed to get all the kinks worked out. Or it can ship badly. Barton shipped badly, and AMD suffered from it. And Barcelona shipped badly, and AMD suffered from it. Side note to self: if AMD code names a CPU Barbarosa, sell them short. And on the flip side, Intel's products in 2007 shipped very well. It took both those things happening to put AMD in this deep of a hole. That's not to say AMD would not be in some kind of a hole anyway, but cut 2-3 quarters out of the depth of hole and I doubt anybody except the most anti-AMD folks here would be talking abotu bankruptcy.

Source two also has some down the road negatives. Although Barcelona's problems were probably only going to hit a very few servers, AMD made the decision (probably wisely) that they weren't going to ship a CPU with a significant known problem in it even if that problem was very hard to reproduce and unlikely to hit people in servers (and never on desktops). Thus, a whole lot of AMD resources that had been supposed to be working on making volume quad core CPUs were forced to find other things to do, and none of them made as much money (on older CPUs) as they would have made on AMD's latest and greatest. What's worse, instead of running production volumes of quad core wafers, which always improves yields and performance in small ways, they were making products on which almost all the improvements had been found. AMD *might* have been better off just making tons of quads, forgetting all about triple cores, and then trashing all the quad cores but at least gaining all the learning curve so that today Barcelonas would be better. But it's really easy in hindsight to say that expectations were insufficiently pessimistic. I *think* AMD's expectations were mildly optimistic, but only mildly. Remember how quickly AMD got their first dual core out? So AMD did not have much reason (before it happened) to expect this bad a result.

That said, I foresee three ways that AMD survives as something other than a bunch of IP plus a major lawsuit victory. None of them are what I'd call highly likely; I was far more optimistic in the dark days before Opteron than I am now. But we do have to recognize that AMD has more customers than ever before and has had very few cancellations because all of those customers know that only AMD stands between them and Intel getting back up to 60% margins on the backs of truly monopolistic pricing.

Way one, which I put at very unlikely, calls for AMD's management to be more accurate in their forecasts than they have been for two years or so. They might just manage to trim the right costs and catch up in sales to get back in the black this year. Barcelona could be astonishingly good. Intel might develop hiccups allowing AMD to catch up. I will say this: if AMD does not break even in Q4 this year, I expect Ruiz to become an ex-CEO. A steady diet of unfulfilled forecasts is far worse than a diet of less optimistic but met goals.

Way two, which I think could happen, is for Fusion to work really well. I'm not talking about what they are shipping now; I mean the truly integrated products with CPU and GPU sharing memory resources as needed for best performance. The potential gains (if that all works out) are significant: cooler running products (no fans, lower power, lower TCO) where the performance gains from integration compensate for (at least mostly) CPU deficiencies. I don't really expect those to arrive until next year. If they do work, then AMD would have a substantial cost advantage in volume desktops and low power notebooks, and since those would be reasonably small to manufacture AMD would make a lot of money. Make no mistake, AMD overpaid for ATI, and has taken the write offs to reflect this, but if AMD did not own ATI today I would see almost no chance of company survival. If the performance gains are sufficient, AMD could also catch and pass Intel for gamers, but I think that is unrealistically optimistic. I'm sure Intel hasn't been sitting around just letting AMD catch up.

Way three, and this is a longshot, does come from acquisitions. I have no real idea what strategic goals Abu Dhabi has from their AMD investment, but if they really want to own a high tech company, they can guarantee AMD stays in business. I'm pretty sure that the reason AMD is nearer $6 than $4 a share right now is the Abu Dhabi capital. More likely to me (and this is just a thought) is that IBM may decide to get back into CPUs to better leverage their new string memory. Imagine putting ten times the cache memory plus RAM onto a CPU/GPU in a shared Fusion environment with a CPU that takes up less space than just the CPU does now. IBM is only making cell processors right now, and they are very specifically tasked. I think it possible that if IBM's string memory pans out as they say that IBM will decide to start making CPUs again through acquiring AMD. It would have to be far easier to integrate that memory into AMD than Intel CPUs, and it would be easier for IBM to buy AMD than to start designing general purpose CPUs again.

Given that AMD did reduce their losses, is finally shipping Barcelona, and has not lost any customers, I think it not impossible that AMD will return to break even in Q4 as they forecast; not likely, but not impossible. If Fusion works out, AMD could well make money in 2009 and for a while thereafter. But I would call any investment in AMD right now highly speculative.