Tag Archives: amd intel chips semiconductors markets

AMD reports tonight, how bad will it be?

Poor AMD, I’ve been following and trading in this company for nearly a decade. If there’s one thing that always (has) been true about this company it’s that they always come back – and then blow it all away again.

AMD is down a staggering %85 from their highs two years ago when the Athlon64 was king of the world and INTC couldn’t seem to do anything right. Since then it’s been all the reverse. Intel finally took the worst CPU architecture they ever designed, the Pentium4 off the street and shot it.

In the last 2 years Intel has been executing like clockwork with a blitzkrieg of “Core” architectural releases, tick, tock every six months frequencies go up a notch, power comes down, performance per cycle goes up, cache and number of cores is doubled and so on.

AMD meanwhile has gotten the wrong end of almost every single strategic decision they have made in the meantime. AMD switched sockets to AM2 and new memory type DDR2 in a move that offered no performance increase and orphaned the upgrade path of millions of installed machines (2 years later, enthusiasts with the old s349 socket are upgrading en mass to intel machines as they have to replace their motherboards or the whole system anyway).

AMD’s process engineers failed on the transition to 65nm designs, yielding chips that were slower and with little power advantages. To this day, AMD’s fasted chips are still built on their venerable, but ancient, 90nm process while Intel’s latest are build on 45nm a full two process generations ahead (as with the venerable Moore’s law, each process generation allows roughly a doubling of transistor density at the same cost of production so you can see AMD’s disadvantage here).

Then AMD paid waaaay too much for an acquisition of Canada’s ATI, paying 5 billion, in cash, at a time when AMD’s balance sheet was already strained. Stockholders are waiting for massive writedowns on this purchase to be announced today. Unfortunately the crippling interest charges on the outstanding debt AMD wrote to finance the purchase is not so easily swept away.

Then AMD screwed up Barcelona/Phenom. Phenom was to AMD’s coup de grace for Intel for late 2006, a truly native quad-core processor with an architecture Intel couldn’t match. It turns out, they gambled on an architecture more advanced than was within their reach. And then it was delayed 6 months. and expectations for launch frequencies were managed downwards. And then it was delayed again and frequencies were going to be a little lower. And then it was launched in Sept 2007. in limited quantities. Then the mainstream parts were delayed another quarter. Then the fasted chip was withdrawn the night before the launch. Then a bug was found in the silicon. Manufacturing cycles being what they are, AMD expects fully working 65nm to be back out sometime midway through this year at 2.5GHz, with faster 45nm parts sometime by 2009.

AMD’s market cap is now near 3.8B, the whole company now worth much less than it paid for ATI just a year and half ago. Though, ironically, the ATI component could be the one brightspot in the results tonight.

Somehow though, they will bounce back. Without AMD, INTC has a total monopoly on the chip business, and no one, not even INTC (for the regulatory scrutiny it might attract) wants that. I predict that someone will come to the rescue.

And when they do, replacing the CEO would be a good start. Shareholders should be braying for Hector’s head. Meanwhile as far as solving for the macro strategy ahead, a simple rubrick WWHD-LDWTEO (what would hector do, lets do whatever the exact opposite) wouldn’t be a bad way to start.

disclosure: I am currently long Intel and neither long nor short AMD. Both stocks are trading near 52-week lows.