AMD: Spinning Off Or Being Spun?

Will AMD spin off its fabs soon?  It won’t be that simple.

A few days ago, one stock analyst predicted that AMD would spin off its fabs in one way or another within a few weeks, with the transaction likely to be fueled by Middle Eastern investment money.

Let’s look at the facts in this situation:

AMD currently has two fabs in Dresden, Fab36 and Fab30.

Fab36 makes all of AMD’s CPU, 65nm ones now, 45nm one later.

Fab30 is currently dormant.  Before it was mothballed, it could make 90nm processors using 200nm wafers.

AMD has been selling Fab30’s equipment to a Russian company.  You likely know that it sold 130nm fab equipment to them a while back.  You likely don’t know that it recently sold its 200nm wafer equipment to the same company.  (This company would also like to buy AMD’s 90nm fab equipment, too, and is seeking the necessary technology transfer licenses, but given events in Georgia, this may not be forthcoming soon or at all.)

AMD has said it has been “converting” Fab30 over to 300nm wafer production, but funding for this seems to have been put on the back burner, with priority being given to 45nm conversion at Fab36.

Given points four and five, it is safe to say that unless the wafer sale is voided, Fab30 would require  significant additional investment to make any kind of chips at all, and a good deal more money to make advanced chips.
AMD has a tentative agreement to build a new fab in New York State.  To get the billion dollars in subsidies and other financial incentives New York State promised it, they must commit to building by next July.  If the past is any prediction of the future, it will take AMD about two years to build the fab and get the first CPUs from it, and about two more years to get it to full production.  So AMD will need Dresden CPUs for quite some time to come no matter what it does.

AMD does not now have the money to build a new fab.  Under current financial conditions both inside and outside AMD, they cannot realistically raise the needed money through sale of its stock, nor borrow the few billion needed to build it.  Due to a patent-swapping agreement with Intel, it cannot sell itself to someone with the necessary money without Intel’s effective approval now or for the next few years.  A new agreement (or extension) will eventually be needed, but not for a few more years.

The only realistic option AMD has to finance the fab is to somehow monetize its major assets without breaking the agreement with Intel.  AMD’s major assets are its intellectual property and its fabs.  To make a long story short, getting quick cash from the fabs is a lot easier (and legal) than selling its IP, and probably is AMD’s only realistic chance to free up enough money to finance the New York State.

It’s very easy to see why AMD would want to effectively sell the Dresden fabs.  It is not so easy to see why someone else would want to buy (or at least pay a whole lot for) them.  Fab 30 is essentially stripped, and would take at best a couple hundred million, at worst maybe a billion dollars to get back into operation.

Fab 36 is up and running, but AMD has to use at least most of that fab’s capacity to make AMD chips for at least the next three years.

Right now, even if you’re kind, CPU manufacture at AMD is at best a break-even proposition.  What kind of price would any new entity charge AMD for making processors?  It’s a zero-sum game; if the fabs charge AMD a lot, AMD won’t make (or even lose) money on CPUs, even if the revenue picture brightens considerably.  If the fabs charge AMD only a little, the fab company will lose money, and why would you invest billions of dollars just to lose more money hand over fist?

Besides, the fabs are just tools.   What makes AMD go, what makes it important, what makes it potentially profitable isn’t the tools.  You don’t pay more for a CPU mainly because of the tools made to use them, you pay more for the ideas embedded in them.  In today’s world, creating a product is a lot more profitable than making it. 

For three or four years, Fab 36 will have to make AMD CPUs.  Then what?  Will AMD then abandon it and make all their chips in New York?   Unless AMD has an amazing growth in marketshare, that is what is likely to happen.  What happens to Fab 36 (which will certainly need an expensive retooling by 2012) then?  They’ll go from enslavement to abandonment.

Would you want to pay three or more billion dollars, then billions more for retooling and upgrades, just to be AMD’s handmaiden?  Only if you were born yesterday.

All this does not mean a deal is impossible.  What it does mean is that any deal is going to be a lot, lot more complicated than just spinning off Dresden fabs into a new company and have some Arabs pay billions of dollars for a deal that will leave the Arabs with all of the burdens and AMD with all of the profits.  That is just fantasy (and it may be quite possible that’s all AMD is doing).

If there is a deal, rest assured that whoever buys into the deal is buying AMD or at least the option to do so later on.  Maybe not technically, maybe not according to the strict letter of the law, and believe me, they’ll obscure the reality any possible way they can deep in the fine print of any agreement, but that is what is going to happen.  They probably will let Dirk or Hector or whomever front the company and do the day-to-day stuff, probably give them a pretty free hand in doing it.  But when push comes to shove, they’ll have the final say.

What will this mean for AMD?  At least in the short-term, nothing but good.  AMD will have found its sugar daddy.  I don’t think even the New York State fab will be thrown under the bus as a result of this, the fab will get built, and the new owners will own a major stake in that, too; it would be near-suicidal for them not to, for both political and economic reasons.   In the long run, well, maybe Fab 48 will be built on the Arabian Peninsula.

Could I be wrong?  Sure.  Those Arab investors could have been born yesterday, or AMD is negotiating with them like they were.   I doubt the first, and there isn’t going to be a deal if it’s the second.

To summarize and conclude, if this happens, this deal is going to get sprung on us, it’s going to be very complicated, very detailed, and the details will be very obscured.  It may well be presented as something simple and straightforward, but it won’t be.   It may even look like a relatively minor deal at first glance, but if that’s the case, there will be plenty of clauses and options to change that later on when people aren’t looking.   But if this happens, the die is cast, and AMD will effectively, if not technically change ownership.





About Ed Stroligo 95 Articles
Ed Stroligo was one of the founders of in 1998. He wrote hundreds of editorials analyzing the tech industry and computer hardware. After 10+ years of contributing, Ed retired from writing in 2009.

Be the first to comment

Leave a Reply