AMD’s Greatest Gift to Intel
by Rick C. Hodgin posted on February 24, 2003 11:19 am
AMD didn’t want to give Intel any gifts, but it has. What could it be?
“There you are.” The words spoken by AMD as it hands Intel the future of x86 computing. Intel’s reply: “We knew you’d come around. Just a matter of time, really.”
I have been thinking about the AMD/Intel situation and how much it has changed in the past 12 months. A year ago we were expecting Hammer by the end of the year. A year ago AMD was on track for Hammer while producing Athlon XPs that competed admirably with Northwood P4s (albeit somewhat less than with Willamette P4s). A year ago the future seemed bright for AMD, and the possibility of x86-64 becoming mainstream was almost a certainty. A year later and oh, how the mighty hath fallen.
I’ll say it simply: AMD has handed Intel the future of x86 computing. Without performance numbers nearly doubling that of a comparably priced Pentium 4, AMD is finished as a top-end competitor to Intel. It’ll be pushed back to an extremely distant second place, and will likely hold less than 5% of the global marketshare in the years to come.
As much as I hate to admit it (as I still believe very strongly in AMD’s business and product goals), I have to. I am forced to. The situation reveals itself to us all, and that’s exactly what it’s saying. Why, you ask? Because the performance is no longer there. The price advantage is no longer there. And the product association AMD enjoyed through 2001 and into early 2002 as “better than the price would indicate performance” is also no longer there.
AMD is now known as a company that delays and stalls and releases paper products, and when it finally does release new products they don’t perform as well as expected. It’s such a change from the Athlon XP image of old, and it’s such a shame because AMD really had a good thing going. It’s done the Athlon XP line a great disservice. So much so that I personally believe the recent K7 products should’ve been given a different name.
Unbeknownst to most all of us (and to AMD also, I’d wager), it now seems that AMD and Intel worked together throughout 2001 and 2002 to put an end to their x86 business. They did it by bringing out faster and faster products at lower and lower prices, and did so at an unsustainable rate. In early 2001 a 1GHz processor sold for around US$1.00 per MHz. Today it’s around US$0.20 per MHz–more than double the performance at around half the price. That curve could not be continued forever, and it seems now that only the biggest boy on the block (Intel) will be able to weather the storm by continuing to release faster and faster products in a timely manner.
I see no way for AMD to regain its previous position based on its current roadmap. The company’s products have been delayed for too long to be high-end performers when they finally hit the store windows. When they are released they will be moderate- to low-performers relative to the fastest P4 of the day, but with one plus: 64-bit ability. And, unless something changes dramatically, it is unlikely there will be much need for the average user to go out and get a 64-bit computer. The average user doesn’t need that much addressing space or that much computing power. And since the price
erformance ratio will no longer be there, what will be the point? Average users will go to Dell or Gateway and buy the $599 special with a Celeron 1.8GHz processor and be done with it.
The only way I can see AMD regaining their previous position is if it is delaying Hammer right now to re-engineer the core to include some ability or feature that will allow the company to produce a product come September which provides 1.5x (or greater) performance relative to the Pentium 4 of the day. Without that kind of powerful, high-end product, I feel AMD will be stuck forever in second place with more than a 90% marketshare between itself and Intel.
In closing I’ll leave you with a home-made joke written by yours truly. It is a sad bit of commentary on the state of affairs at AMD and is admittedly not too funny … but it is proving to be very accurate. Be sure to tell it to your friends. How many AMD engineers does it take to build a Hammer? More than AMD has on staff today.
