AMDINVESTOR.COM

GPU+CPU combined on one die together by 2011

http://en.expreview.com/2009/04/09/amd-sabine-platform-to-support-directx-11-in-2011.html

AMD has recently been rising, moving up over 50% in the span of a month.  As seen in the graph below, AMD is the blue graph whereas the other graphs are the general market.  There could be many reasons for this, but I think the market is beginning to see that AMD has a lot of upside with its ongoing re-structuring.

1, Out of the fab game

For one, AMD is now moving out of the fab game.  It’s an expensive game to be playing, one in which you need state governments to help you keep up to date with the competition.  Consider AMD’s fab play in Dresden or New York, in both cases you need state support to run those expensive fabs.  Why is the optimism coming now?  Because it initially did not look like it would go through, but voila, it has, and AMD has some debt and future fab worries off it’s shoulders.

2.  Fusion still happening

Although it’s true some people have been exiting AMD, overall the AMD team is still busy on fusion.  It’s not a simple thing, for example, putting a GPU and CPU on the same die creates a lot of heat issues.  But the team is still steadily working on the GPU+CPU dream.  If you think of it, AMD now is no longer a CPU chip maker, but a designer of platforms.  And they happen to have the palette of CPU, GPU and Chipset to choose from, something very unique to them.

I haven’t written in the blog for a while, sorry for the delay.

I think we have all been through quite a nauseating journey watching AMD stock consistently slide.  I would like to say that I know AMD has bottomed, but there’s no certainty to that.  But I can say, you are probably better off buying AMD now at 2 dollars compared to my average price of 6.

So here’s my speculative view on AMDs situation.  It’s a smart company with lots of smart people, but it’s stuck in a tough spot.

First of all, the semiconductor industry is slowing down.  When you think of it, applications don’t require faster processors.  When’s the last time you fully utilized two cores?  Maybe this will change one day, but for the *average user*, extra cores, and extra clock speed no longer matters.  Hence the growth in net books.

Second of all, AMD wants to do the fusion thing, bringing CPU and GPU together.  But as we are seeing now, it’s actually harder than expected.  In fact the project has been delayed.  The truth is that CPUs and GPUs consist of millions of transistors, and integrating the two architectures is easier said than done.  The advantage of such a system, is that you can cut silicon with both parts on one slab, and so you save money/power.  But really, is that necessary in this market?  It’s not that there isn’t a return, the question is if it’s worth the investment.

The last point is that if something saves AMD, it won’t be the new emerging markets for their chips, or a new processor called fusion.  It will be the successful execution of their fab lite strategy, and the success of the foundry in shipping out chips, while leaving AMD to the design.   Also, AMD can be more flexible choosing various vendors for their chip manufacturing.  This OPEN approach, is kinda like linux development in that it makes AMDs development approach more flexible.  But I would argue to AMD, find ways of outsourcing your design as well, not just your manufacturing, not all your design has to be done in sunnyvale.

It’s nice to know in this climate of financial instability, that AMD is giving us the sense that it is beginning to turn the ship around.  AMD has released a new ad campaign geared around the “fusion” concept they used with their GPU and CPU.  As an idea, this is great, but as always, AMD will need to prove itself in its execution.  Let’s hope our boy Dirk will prove himself.

A recent article in fortune provides a discussion with Dirk Meyer over Asset Smart and shows the first signs that he is confirming it is going to happen for sure this coming year:

  http://bigtech.blogs.fortune.cnn.com/2008/09/08/re-engineering-amd/

I just picked up some AMD again because it seemed to be a good deal.  Admittedly with high gas prices, lots of stocks are trading low, but I think that AMD is facing a turn for the better.  AMD has released a series of highly popular graphics cards at competitive prices and also it is offering terrific PUMA platform for notebook computers.  We’ll see how this goes.

AMD dropped today and the day before, going from $8 to  $6.80.  The exact reason no one could be sure of, it could be recession scares, but it probably was due to Nvidia’s acquisition of physics processor maker AGEIA.  Personally, I do not see why this purchase should affect the long term fundamentals of AMD as a company.  AMD has long said that it is no longer just chasing the performance crown, like wal-street would want.  Of course, physics acceleration is an important thing, but it’s not the only thing.  Power consumption, performance-per-watt are equally important, and those things can only be solved by a platform approach, which Nvidia is ill-positioned for.  Even if this is a performance advantage, it will take some time before AGEIA technology makes its way to the GPU and by that time, AMD would have moved ahead some distance.

From a purely speculative perspective, AMDs 52 week low was $5.30, and it is trading about $1.5 above that, approximately 30% above its low.   The Price-to-book ratio is less than 1.5.   It’s still near the bottom and still an excellent opportunity for buying in.

If you noticed recent news, there’s been a tirade of coverage regarding the recession that may happen or may be happening.  Most stocks have been seeing red, but surprisingly, AMD has been doing fine, holding its position.  Intel has also seen some recovery.   AMD looks to be on the road to recovery, while other companies are just beginning to suffer losses.  Even stocks like Apple and Google are suffering heavy casualties in this environment.    I would suggest people look into AMD and evaluate whether or not it’s a buy.  Personally, I believe AMD at a $5 range is a pretty good deal, however even at $7 (which I’ve moved in on) it’s still a reasonable deal.

I’ve also found  this article by an Intel manager about the kind of PC growth being seen in China as rather interesting.  The “Art of the Price War” report I’ve found there to be equally insightful into how chinese companies compete price wise, just as Intel has used the same tool on AMD.

AMD will be announcing their earnings, and providing a detailed report on Jan. 17.  My only advice to all those out there deciding whether or not to invest in AMD, please, please, don’t depend on second hand information.  There are lots of spinsters out there, the financial media, analysts, and really, none of them can provide you a true picture of AMDs financial health (or sickness) through limited sound bites.  Don’t buy into sensationalist headlines, because if you do, you’ll find yourself short sooner or later.   I remember reading about AMD’s analyst day, and how it was such a gigantic disaster.  Every analyst had something negative to say about the event.  However, if you bothered to actually visit AMD’s investor relation site you would get a chance to hear and see the video event yourself.  What you would have seen, is that things are not as black and white as they seem.  There are positive things at AMD, and the absolute pessimism is undeserved.   For some reason though, AMD has taken down this valuable link, perhaps because the coverage of their analyst day was so negative.   If you are interested in pricing AMD accurately, don’t rely on the word of people who’s interest it is to get into the headlines, to pull on your heart strings, and motivate you to behave irrationally (either buying or selling).  Just as AMD may have been a shade to high when it was in the 40s, pumped up by analysts, it might be a shade too low now, as those same analysts dump on it.  Don’t give these analysts too much attention, you are making their careers, instead, rely upon your own analytical faculty and listen to them as only an alternative perspective.

I’ve seen a lot of posts that suggest that AMD should switch over to an “Asset Lite” model by outsourcing most of its manufacturing ability, in fact some seem to suggest switching to “Asset Lite” would mean a major upside for the AMD company.   However, in my opinion, such a move would be a mistake because there are many advantages in keeping fabrication in house.  No matter how much TSMC, or other fabs for hire may innovate their processes, they can not keep up with Intel’s innovation prowess.  The recent introduction of Hafnium based Hi-k silicon technology allows Intel to “leap ahead” of their competition by 2 years or so, with fundamental technology that cost effectively reduces gate leakage up to 10 x, and power by 20%.  Intel will be able to reduce die size faster and faster compared to AMD.  So in order for AMD to  compete, it needs to invest in fundamental technology as well to catch up with Intel’s lead, such as when AMD and IBM announced Low K Dielectric in future 45nm ICs.

Being a CPU manufacturer, naturally means this competition in fundamental technologies.  Outsourcing to contract fabs in Taiwan will not bridge this gap.  Perhaps that’s why AMD CEO Hector Ruiz described an “asset smart” or “hybrid model”, where some things were outsourced (perhaps graphics chips), and others remained products of AMD fabs (cpu chips).

Moving to an “asset lite” model is definitely not “asset smart”.  For one, in doing so, it’s giving up the “fundamental technology” race with Intel, and forcing AMD into a “design shop” only.  What creates the best performance?  In some cases, with the Athlon processor, it depends on great design.  In others, like the Peryn processors, and to some extent the current Intel Quads, what they lack in design, they more than make up in fundamental transistor technology.  That’s why AMD can’t give up either race, it’s not certain which way is the best.

So, now that Intel has the hafnium technology, is AMD screwed?  I don’t think we can jump to that conclusion.  Because as Intel moves toward a successful CPU company based on solid fundamentals, AMD moves towards a platform company.   What saves overall system power more:  i) a Intel hafnium based CPU that  leaks less current, or ii)  a well designed AMD platform of CPU, GPU, and chipsets that intelligently talk to one another to efficiently reduce power?