Intel Corp (INTC)

[SemiconEng]

AMD hints at 90nm ship slip

http://www.theregister.co.uk/2004/05/18/amd_90nm_slip/
An AMD executive has tacitly confirmed that AMD will not ship 90nm Opterons in significant volumes until the fourth quarter.
Quoted in the company's announcement of its Opteron 150, 250 and 850 processors today, AMD's Microprocessor Business Unit chief, Marty Seyer, says: "We have already begun initial production of 90nm AMD64 processors, and we are on target to begin shipping 90nm processors for revenue in the third quarter."

The key phrase is the reference to "shipping... for revenue". That's essentially the same language used by Intel last year when its said it was on track to make "revenue shipments" of its 90nm Pentium 4, 'Prescott', by the end of Q4 2003. In the end, Prescott was not launched until 15 February 2004.


*** See, I'm not as confident as some in AMD's ability to get 90nm into "High Volume" Production. In the semiconductor development work that I've done, during the development, each engineer is very compartmentalized. They're mostly concerned with making their specific module "work". The problem with that, is that you mostly won't see the "integration" issues between the modules until you start to ramp. Then all those inter-module issues come up.

*** The second issue to bite people, is that the initial engineering lots are almost always run on the best tool in the individual module. while that is great for seeing if the process works, once again, you are not going to see the influence from tool to tool variation until you start to try to ramp.

*** Of course, eventualy you work out all the issues, the question is, what is the magnitude of the integration issues, and how long will it take to fix?