Re: I have a feeling this will happen when the hammer arrives. It will emerge far larger than than it seemed at first and will instantly be the big kid on the block.
The behavior of third parties like Asus, VIA, NVidia, etc. make it appear to be very likely. AMD could still have trouble ramping full production volume, but the samples they show behind closed doors must be very convincing.
Meanwhile, switching Austin to flash cuts AMD's costs by about a quarter, while the move to the smaller thoroughbred should keep units at about the same level. So, if AMD can keep ASPs from dropping more than 25% they'll do OK (so they'll need to maintain about $65 ASPs in Q3).
With total units for this quarter and next not looking to increase, Intel has to be very careful that it doesn't ship too much product and crash ASPs. That should provide something of a pricing umbrella for AMD.
Given the limited number of units the market is demanding this quarter, and probably next, Intel must price its processors to yield an ASP of about $156.
Intel gets about 80% of its revenue from CPUs, and has quarterly costs of about $5.5 Billion which means they need $4.4 Billion just to achieve pro forma breakeven. Their minimum CPU revenue is probably more like $5 billion to keep the stock from completely collapsing. With about 32 million unit sales expected, they need an average of $156 per chip. We know they've been having trouble with the market avoiding all but the lower priced parts, so that limits how low they can price parts that could compete with Athlon 2000+ and higher. It doesn't really matter whether or not Intel could flood the market with P4 2.66 chips or not, next quarter. If Intel could supply all demand with 2.5ghz chips and price those chips at $100, they could knock AMD down to about 5% of the market, but Intel would find itself losing $1.5 to $2.0 Billion for the quarter.
If total unit demand were higher, I don't doubt that Intel could accommodate it, and then they'd be able to sell their chips for a lot less and get the total revenue that they need. But unit demand is down, which means Intel must keep ASPs high, which puts a pricing umbrella over AMD's ASPs. All this raving by Elmer, BMW, and Monica about wafer yields and cost per die overlooks the degree to which costs are fixed in the CPU biz. Last year, you built your FABs and hired and trained your people that will be making chips this year. If Intel doubled or halved their unit production it would probably change their costs by about 10% to 15%, up or down. They need X dollars of revenue from CPUs each quarter, almost regardless of the number of parts they produce. When unit demand is down, ASPs must be kept high. As usual, Jerry saw it coming, and AMD has been, almost frantically, cutting costs for more than a year - while Intel raised its cost base with a huge FAB expansion program.
AMD is in particularly good shape because it now has products in the desktop, low power desktop, desktop replacement mobile, thin & light mobile, workstation, and server segments. In the past, Intel would bomb ASPs in the segment or two in which AMD was competing, and make up the total revenue loss in the segments in which its monopoly was intact. Those days are gone.
In Q4, hammer arrives, mirror bit volume will be at high volume, and AMD will be able to command higher ASPs for both CPUs and flash with its compelling 64-bit CPUs and faster, more reliable flash - but AMD's costs will continue to be very low. AMD is well positioned to make good money in Q4 and mint money in 2003 and H1, 2004. By H2, 2004 or 2005 Intel will also have an SOI process and will be competitive with AMD once again - but, by then, the position of the two companies in the industry will likely have permanently changed, with AMD being as large and wealthy as Intel, if not larger and wealthier.
Certainly, AMD could run into horrible problems ramping Hammer or mirror-bit, but both products are now into the production sampling stage, (shipping, in the case of mirror-bit) and signs from third parties look very, very, encouraging (as you pointed out).
Intel looks like it's in the position of 70% market share, no one can ever touch us, GM, just before Toyota, Honda, and Nissan began ramping up their production. |
