By: Doug S (foo.delete@this.bar.bar), January 30, 2013 9:50 pm
Room: Moderated Discussions
Mark Roulo (nothanks.delete@this.xxx.com) on January 30, 2013 3:51 pm wrote:
> Unless we think that these customers need the RAS (including ECC), the relevant ASP might be
> the Value Desktop chips at $57.92. I'm guessing that these are single socket, quad core, dual
> memory channel chips. I'd expect the individual cores to be faster than current ARM cores, so
Can you buy those value desktop chips on a server motherboard from anyone? Buyers aren't going to want to crowd a bunch of desktop towers into their racks, they want something like that to fit in a 1U rack in their datacenter with manageability functions they come to expect from server products. I mean, technically it is possible for someone to offer this as a server product, but is it this product available anywhere today or only a theoretical possibility? Intel isn't encouraging that now, because they want to sell you server parts at a higher ASP and margin. The fact they COULD go lower indicates there is some slack in the market for ARM OEMs to try to exploit.
Obviously Intel can make life much harder for them by encouraging those desktop parts to be marketed for servers, but that still damages Intel's profitability when compared to the current scenario where those potential $57 ASP customers are forced to buy a $252 ASP product instead despite it being essentially the same thing. Intel fought this battle before with AMD, who nearly went bankrupt in the mid 90s and then adopted a strategy to compete on the low end and had enough success they forced Intel to introduce the Celeron.
The big difference here is that Intel holds the cards for x86, only AMD has a license to make them. They keep them around, but keep them down and that has succeeded as a strategy for a couple decades with the exception of a few years when Intel went down the P4 rat hole and let AMD create x86-64 against Intel's will. They can't use this strategy against ARM, since it is available for any company to license, so it would be like trying to play whack-a-mole keeping them down, if they stop one there will always be another. They can keep them in check, but that may require accepting permanently reduced profits on a certain segment of the market. If that segment of the market is large enough, Intel will really suffer in the future. Thus we see another reason to pursue foundry business while they still have a process advantage over the competition. If they wait too long and their profitability was damaged, they might not be able to make the investments necessary to maintain that advantage, and there's little chance they'd be able to compete with TSMC for foundry business without their process advantage.
> Unless we think that these customers need the RAS (including ECC), the relevant ASP might be
> the Value Desktop chips at $57.92. I'm guessing that these are single socket, quad core, dual
> memory channel chips. I'd expect the individual cores to be faster than current ARM cores, so
Can you buy those value desktop chips on a server motherboard from anyone? Buyers aren't going to want to crowd a bunch of desktop towers into their racks, they want something like that to fit in a 1U rack in their datacenter with manageability functions they come to expect from server products. I mean, technically it is possible for someone to offer this as a server product, but is it this product available anywhere today or only a theoretical possibility? Intel isn't encouraging that now, because they want to sell you server parts at a higher ASP and margin. The fact they COULD go lower indicates there is some slack in the market for ARM OEMs to try to exploit.
Obviously Intel can make life much harder for them by encouraging those desktop parts to be marketed for servers, but that still damages Intel's profitability when compared to the current scenario where those potential $57 ASP customers are forced to buy a $252 ASP product instead despite it being essentially the same thing. Intel fought this battle before with AMD, who nearly went bankrupt in the mid 90s and then adopted a strategy to compete on the low end and had enough success they forced Intel to introduce the Celeron.
The big difference here is that Intel holds the cards for x86, only AMD has a license to make them. They keep them around, but keep them down and that has succeeded as a strategy for a couple decades with the exception of a few years when Intel went down the P4 rat hole and let AMD create x86-64 against Intel's will. They can't use this strategy against ARM, since it is available for any company to license, so it would be like trying to play whack-a-mole keeping them down, if they stop one there will always be another. They can keep them in check, but that may require accepting permanently reduced profits on a certain segment of the market. If that segment of the market is large enough, Intel will really suffer in the future. Thus we see another reason to pursue foundry business while they still have a process advantage over the competition. If they wait too long and their profitability was damaged, they might not be able to make the investments necessary to maintain that advantage, and there's little chance they'd be able to compete with TSMC for foundry business without their process advantage.