AMD Posts $600M Loss On Weaker Sales; SeaMicro Purchase, Foundry Deal

AMD's burgeoning profitability took a hammering in Q1 thanks to a $700M deal with GlobalFoundries, its purchase of SeaMicro, and a sequential decline in sales. Total company revenue was $1.59B (down from $1.69B in Q1 2011), but Sunnyvale delivered a $590M net loss. Much of that is attributable to the company's new deal with GlobalFoundries, in which AMD paid GF $700M, and transferred all of its ownership stake in the company in exchange for "non-exclusivity" for 28nm APU production.

Put simply, AMD paid GF for the right to manufacture 28nm APUs somewhere else -- and they paid through the nose.

Computing solutions revenue fell 6 percent sequentially while graphics revenue held constant. One positive step for graphics was an increase in operating income; graphics earned AMD $34 million on sales of $382M. That's a meager 9 percent operating margin, but it's a big step forward from Q1 2011, when the company reported OI of just $19M on sales of $413M.

The GlobalFoundries deal and Seamicro purchase guaranteed that AMD's financials for Q1 would be pretty lousy; the actual figures aren't too surprising. Rory Read noted that Best Buy's #1 laptop SKU in Q1 was an AMD product (not that this means what it used to), and that APUs now account for nearly 100% of the company's mobile shipments.


Server ASPs were down, despite Bulldozer's launch in late 2011, likely reflecting the fact that AMD has a great deal of trouble competing with Intel in this space. The good news, according to Read, is that the company has turned a corner with regard to 32nm availability and expects to ramp products like Trinity much more quickly than Llano managed in 2011. Read also teased at a headwind from "quite a significant number of 28-nanometer products that are going to start to ramp in the second quarter." AMD's new 28nm mobile products are set to debut in Q2 alongside a revamped Brazos 2.0, but we expect virtually nothing from the latter. The so-called "Brazos 2.0" is Brazos 1.0 with a new name and very slightly tweaked clock speeds. With Brazos stuck on 40nm until 2013, there's not much AMD can do to improve it.

An Interesting Exchange

There was one exchange of particular note in the Q&A section of the conference call. At one point, Seifert said: "And we will see, in the second half, however, also some more headwinds with 28-nanometer products going in production at GLOBALFOUNDRIES. That those have control impact in the second half."

This lead to a follow-up question, in which analysts asked AMD if the 28nm production plans at GF were strictly GPUs or if APUs were part of the mix.

Seifert: They are APU products. Yes.

David Wong: They are APU products? 28 -- so you will be, actually, you expect to be selling 28-nanometer APUs in the second half that are made at GLOBALFOUNDRIES?

Read: No. No. No.

Seifert: From a volume perspective, very little volume. But we start the manufacturing ramp in the second half so we can get ready for the launches in 2013.

What makes this so interesting is that AMD paid GlobalFoundries $700M and gave up its ownership stake in the company precisely so it could build APUs elsewhere. This is the first time AMD has gone on the record stating that it plans to do any 28nm production at GlobalFoundries, but note that it takes a very indirect route to saying so. All AMD is willing to admit is that some APUs will be ramped at GF in the future, with zero reference to which products or when those products will launch.

Once we toss out the one-time costs that nuked AMD's earnings, the company's position is reasonably solid in the short term. Buying SeaMicro, paying off GlobalFoundries, and the number of prominent employees leaving for other pastures are all examples of how AMD's plans for itself have evolved and changed in the past nine months. It's still too early to call the results of those changes; we'll be in a better position to judge Read's effectiveness 6-9 months from now.