Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: TGPTNDR who wrote (157622)	2/3/2002 11:56:30 AM
From: Elmer	Read Replies (1) \| Respond to of 186894

When I'm looking at COGS/30M ( Giving Intel credit for selling around 30M CPUS/Q -- I think that's more than they're actually selling, but other opinions will think it's less.) we're looking at a cpu cost of around $100 back in Q1 01 rising to $140 in Q4 '02. Sounds expensive. Maybe I missed something too. Are these numbers from the processor group only? You know that Intel sells far more non-processors than processors. EP

To: TGPTNDR who wrote (157622)	2/3/2002 11:59:08 AM
From: Tony Viola	Read Replies (1) \| Respond to of 186894

tgptndr, >Interesting posts. Thanks for putting them up. (Table formulated showing COGS, WIP, etc. for Intel in 2000, 2001). Too bad the source of that post, TMF, is going pay per view in a couple weeks or so. What they did do is award some free 1 year memberships based on "loyalty and excellence". That'll keep some of the best and then it will depend on how many other folks spring for the $20(?) yearly membership. I wonder how SI is doing financially? Tony

To: TGPTNDR who wrote (157622)	2/3/2002 12:28:10 PM
From: Dan3	Respond to of 186894

Re: we're looking at a cpu cost of around $100 back in Q1 01 rising to $140 in Q4 '02. Sounds expensive. That's the same number I come up with as an estimate of costs for Intel's average CPU. The interesting thing is, almost all of the cost is due to two factors - capex and administrative/marketing. The actual cost of running the wafers is the least significant cost (other than the fact that FABs are tied up in the process). So that moving from 217mm2 to 145mm2 P4s on wafers that have gone from 200mm to 300mm has almost no affect on CPU costs unless it allows them to reduce capex/administrative costs by building and running fewer FABs. But we know that, in fact, the opposite has been happening. This is why Intel has stated that no savings would be seen this year from the move to 300mm (they forecast some savings could be seen beginning in 2003). There are aspects of Intel's FAB expansion program that just don't make sense. AMD expects to go from 30 million parts out of 1.5 FABs to 50 million parts from 1 FAB (their hope). And they're also expecting their costs to fall. Intel is going from 120 million parts from 4 to 6 FABS to 180 million parts from 6 to 8 FABs (their hope). Despite the shrink to .13 and despite the move to 300mm wafers, there is no way Intel can have anything but higher per CPU costs going forward, given all the FABs they had to build and run - while AMD's costs are falling. Message 16829926 All of Elmer's screeching that AMD's yields had to be poor may be a reflection of what he's been seeing in Intel's plants. From the amount of plant Intel clearly feels it must place on-line, they appear to have concluded that that they will have to push their processes way beyond the sweet spot in terms of yields in order to get parts that will perform competitively. AMD's early move to copper put them in a position where they could dial up performance pretty much whenever they wanted to. Intel has now moved some of its plant to copper, and some even to .13, a quarter ahead of AMD. And P4 on .13 copper looks good for an easy 2.5GHZ. But AMD has been developing an SOI .13 alongside their bulk (Intel style) .13, which should put them back in the position of being able to dial up performance at will by the end of next quarter. Knowing they will have to push a bulk .13 to compete with an SOI .13 may by why Intel is being forced to load itself up with high plant costs. As CPUs become commoditized, and CPU ASPs, in general, continue to fall, Intel stands a good chance of getting squeezed. The cost of replacing someone's PC continues to be much more than just the cost of the box - it's the time to move from one machine to the next that really costs. So buyers will always have an incentive to buy more "CPU", at a higher price, than they appear to reasonably need (the hope being that this will defer the next migration). But this "upgrade insurance" is now available from an Intel competitor, and that continues to depress the price buyers need (or are willing) to pay.

To: TGPTNDR who wrote (157622)	2/5/2002 5:24:04 AM
From: Amy J	Read Replies (1) \| Respond to of 186894

Hi TGPTNDR, RE: "that's 3 1/2 weeks build time for Intel's average chip...What is it?" This old SI post might be useful: Regards, Amy J ----------------------------------------------------------- "how long does it take to manufacture chips? I seem to recall you saying it takes a month or two? " "That all depends on the manufacturer and a zillion variables. In Intel's case, the time for wafer processing was about 6 to 8 weeks - from Wafer Start to Wafer out. Next, the wafers need to be "sorted" - that is, each die site on the wafer is tested for DC parametrics and functionality. Bad devices are are identified at this stage. Sort can take several days to a week or more. Following this, the sorted wafers are sent to assembly - generally off-shore (usually air-freight) - and this operation can be 2 to 4 weeks - depending on "loading" of the assembly house, prioritization, etc. Following assembly, the completed device/module will undergo final testing - in a handler that heast the component up to high temperature (80 - 110 Deg. C) and a full functional test, AC and DC parametric testing, and speed testing is performed. This testing is quick ( several minutes/device) but the overall time consumed may be days - sometimes a week or more - again, depending upon factory loading. Finally, marking and packing and shipping must be performed. All-in-all, three months - start to finish - is quite typical. Variations of a few weeks in either direction are possible. Paul" ---------------------------------------------------------