AMD needs money?

I don't think buying ATI was a mistake per se. However, IMHO it was a case of "too soon, too much". Not to sound like a parrot, but I think they could have saved themselves a lot of money had they waited a couple of quarters, and I am not just saying it with benefit of hindsight.

 

I don't think it would be as bad. They have significantly more debt and a big lead weight in ATI right now. I think their stock price and finances would be higher if they hadn't bought ATI.

Maybe not like that, but I do believe they didn't give nearly half as much thought as they should have. I totally agree with you on AMD getting a smaller company like VIA or SIS. The CPU+GPU isn't going to do anything impressive anyways, so a low-end GPU provider is all that's needed. The $5 billion spent could have made one really cool CPU and a fab or two needed to make that CPU.

 

That could be true too. Getting ATI for $2 billion may have made more sense. Still, that wouldn't arrest their own decline in stock price so maybe they were compelled to buy earlier.

 

You say tomato, I say tomato. Or whatever.

Look at December 2005 when these conversations started and AMD's share price is at pretty much an all-time high, more than 50% higher than it was when the deal was announced later. Since ATI doesn't exist anymore, I'm having trouble finding a share price history for what their stock was doing from Dec 2005 to July 2006.

 

http://ir.ati.com/phoenix.zhtml?c=105421&p=irol-stockquote

http://www.thebulltrader.com/592/atyt-buyout/

 

Well if we look back in time at the aquisition of ATI by AMD many predicted that the results would end up in a similar manner to the current state of affairs. Oh well maybe they will sell ATI to someone else eventually Not that I dislike ATI I have a x1900 here and an x800 in my other box. I just was saddened when AMD bought ATI as a consumer.

 

I'd say AMD bought ATI for two major reasons.

-Diversification. AMD should be worried that Intel may get well ahead of their CPU business. It's a real possibility. In the '90s, AMD had flash RAM as another major seller. They lost that a few years ago because of how cheap flash has gotten. AMD needed to expand and not have the entire company relying on their CPUs.
-Image. Intel's unified platform has got to give them a massive edge over AMD. Intel build's everything for their platform and that brings a level of reliability, performance, and maybe most importantly, accountability. The OEMs only have to deal with Intel for their support/implementation issues. And you have to admit that Intel's chipsets are the best out there for reliability and, usually, performance.

And, I think AMD was thinking in the very long term. Not what we're seeing right now. What we have now is just an inconvenient lull caused by both companys' product development ending at about the same time.

As long as the R600 generation isn't a horrible disaster when it is released, and AMD's new CPUs aren't the same, I don't think there's much to worry about. This move was something I think they just had to do.

Hopefully in the future AMD will have a platform with as much polish as Intel's (in the eyes of the big OEMs). Maybe they can even compete with the largest seller of graphics processors (Intel!)

 

AMD's put out an earnings warning.

They expect to report revenues of 1.225 billion, which at worst is half a billion dollars short of their rosy projections from last year, at best it's over 300 million short.

AMD's also planning to cut costs and increase efficiencies to the tune of 500 million dollars in reduced capital expenditures.

They say they won't be changing their capacity plans, though I wonder where they're going to cut to get that kind of savings if it's not out of the fab plans.


I guess we'll see what they're going to gut on April 19th's call.

 

Seems like a company that could really use $2.5 billion in cash right about now.

EDIT: If you think about it, it's pretty amazing. Through its 37 years of existence, AMD's cumulative earnings are +$308 million. I wonder how much money did their executives earned in salaries, bonuses, option and share grand sales over this timespan...

 

Part of the problem is with Intel: They are not playing as a big friendly giant, they are playing a heavy handed Machiavellian strategy. They see weakness in AMD's position and thus they are going to induce a massive pricewar against them. Their goal is to force AMD to cut most of their capital expenditures, which will force AMD very far behind in tech development. Eventually, AMD will be so far behind they will never be able to catch up, and will then be relegated to their original role of discount x86-clone maker.

So far, it's working.

 

With cost of each next process node increasing superlinearly seems that it might be a matter of months until Intel and a very few companies (AMD not being one of them) will have enough volume to realize ROI from process transition. At this point it might be a time to raise a white flag and go fabless, incurring a huge cost disadvantage in the process.

 

If I was still an AMD shareholder right now, I would be, to put it mildly, upset with Hector and his $16M compensation package. He destroyed company’s credibility and balance sheet, and let's hope that that's not all he ruined. Let's look at sequence of events starting at the end of last year: Rosy $1.6-1.7 billion projection while he was trying to secure additional financing. Shortly thereafter, a an earnings warning dropped the projections to $1.55B, which did not exactly endear him to his financiers. Now, another huge warning, down to $1.255B, due to very severe drop in both average units prices AND unit sales. In other words, an attempt to hold on to market share by giving CPUs away has failed.

How bad are these numbers, really? Well, if we are to compare then to the most optimist projections that Hector was so giddy to feed the WS only a few short months ago, we are talking a rather astounding shortfall of $445 million or 27%. That number alone, however, does not give the full picture of how badly the company has been mismanaged over the last year. Consider this: a year ago, AMD revenues were $1.33B! This 5.7% year-to-year DECLINE in revenue, considering all the events that transpired last year, is, in one word, stunning.

 

How is going fabless a huge cost disadvantage? Note that right now NVIDIA has higher gross margins and lower ASPs and probably uses more silicon real-estate on a unit basis than AMD. And NVIDIA's margins are still climbing - there is a good possibility that the new target will be 50%, which is the long-term goal of AMD.

It seems to me the huge cost disadvantage for AMD is having fabs. How many of their 16,000 employees are needed to operate the fabs? Unless there is a lot of deign and market ineffciency, I would venture at least half. That's on top of the MASSIVE capital equipment costs. Even if AMD were profitable that money would go right back into factories. NVIDIA just puts it in the bank.

TSMC is very efficient and has huge manufacturing scale on par with Intel.

This seems like a major recognition failure by AMD management.

 

AMD has little choice.
Legally, it cannot outsource more than 20% of its x86 production. This is the primary limit on its deal with Chartered.

Any x86 product on a foundry process would also likely be non-competitive at the higher-margin segments due to lower clock speeds and longer design times.

Look at the third x86 designer as an example.

 

That figure looks pretty shocking, but there are plenty of much more shocking numbers when you look at the actual numbers and very, very basic and reliable extrapolations you can make. Much of the revenue shortfall is because of CPUs, not GPUs/CE/Chipsets. I'm fairly confident GPUs did OK in Q1. The decline might be higher than seasonal, which is a bit ironic given Vista's launch this quarter, but that's another problem.

Anyway, when you compare Q406 to Q107 in terms of Computing Solutions revenue, Q1 is likely down about 35%+ even after considering seasonality. And margins will be down too, so worst case, you might see a 50% decline in CPU gross profit. Best case, you'll see a 35-40% decline.

Anyway I got a mini-article coming up at Beyond3D tommorow on AMD's revenue warning, so I probably shouldn't give away my entire analysis here. If I had to put a positive spin on this, though, I would just say this: the ratio of revenue attributable to the former-ATI is increasing. And while I didn't respect ATI's management so much, at least they weren't as complete shit as AMD's management.

I'll be curious to see how the street reacts to AMD's forecast for Q2/Q3 on the 19th of April. In fact, I'll be quite curious to see how rosy AMD is going to try to make those forecasts look, and what they will focus on. My guess is that they'll actually focus on R600/RV6xx more than you'd expect them to normally, but we'll see.

I've seen you mention this a couple of times, but couldn't find any obvious info about it. I'd be VERY interested in a source and some extra detail on that!

 

Arun, I am not sure how well graphics devision is holding its own. If you think about it, the two companies went from $1.97B(?)in combined revenue to $1.255B YtY. Looking just at the product lineup market share numbers on ATI's side of things, I would expect their revenue to be off by full 40%, putting them into ~$380M range. That would put AMD Classic(tm) share at about $875M, about a 31% decline.

This is just my opinion, but I think ATI has bigger share of the blame, revenue-wise, then AMD.

 

Heh, there is a reason why "real man own fabs". You yourself control the process. You are not competing with others for floor space. And, of course, you pocket the healthy margins the fab-farm would demand.

I think a cursory glance at 65nm production will not bear this out.

 

I must be missing something - since when is there any indication that ATI's market share is going down drastically? Q4 was actually up over Q1 for desktops iirc, and I don't really see what dynamics would apply in Q1 that would make them lose significant market share against what they had in Q4! Of course, you could predict that ATI's laptop market share is going down further, but that'd be kind of strange for a quarter just preceding a new cycle, aka Santa Rosa.

 

Only three companies has or is close to having a giga-fab; Intel, Samsung, and TSMC. Everyone else has fabs that can churn out ~25k wafers per month, but these three have a fab that can churn out ~100k wafers per month.

 

Keep in mind that AMD is likely to win an incredible amount of money from Intel. Yesterday in court, Intel stated that they "misplaced/lost" key emails subpoenaed by AMD. They have until the 17th to come up with a believable explanation to the court for this.

I see both a monetary award to AMD and a change of practice for Intel on how it operates.

epic