Intel and Mobile

[hist78]

I don't think the average analyst understands just how important volumes are.

If you look Intel's past earning conference call transcripts, normally those analysts tend to ask easy questions. I think they are worried about asking hard questions may make Intel unhappy and it may jeopardize the relationship between Intel and them.

 

[benb]

Cost, schedule and quality are all primary, volume is more of an outcome than an independent variable.

Of the three primaries, quality is the most valuable, because it is difficult to duplicate. Apple has the quality in mobile and profits flow directly from that. Intel has the quality in datacenters. Intel never really pursued the mobile quality characteristics because it would have detracted from the quality characteristics needed in datacenters. Intel has always been a focused business, rather than a jack of all trades. They deserve credit for trying to stretch their datacenter-orientation in completely the other direction, spanning a huge range of power/performance. Apple and Qualcomm aren't even trying to do that. Perhaps Intel's lack of success will dissuade them from trying.

 

[count]

If you look at the history of how Intel got it's lead in datacenter, it wasn't because they had the best chips at the time. The first Xeon chips were basically pentium chips with some extra cache. But they had the volumes from the PC market, and because of that they were able to catch up quickly and nobody else could keep up.

My opinion is volume drives innovation. You can think of any product as having "material" and "information" components in the cost structure. Something like software is nearly 100% information, and something like coal is mostly material. I would argue that semiconductors are more "information" than they are "material". With information costs, they are only incurred once and the unit cost of information in a product is that cost divided by the quantity sold. So if you don't have the volumes to support large informational costs in your product, you aren't going to be able to sell your product at a competitive price.

I actually think this is the mechanism by which Clay Christensen's disruptive innovation works. The nascent firm seizes a growing, low cost market, and growth in that market allows them to innovate at a rate that the incumbent can't afford.

Another interesting data point... Moore's law, which generally predicts technological progress to be a function of time, is actually a worse fit than Wright's law, which generally predicts technological progress to be a function of cumulative volume produced - even for the semiconductor industry. In fact, Moore's law can be derived from Wright's law if you assume demand increases at an exponential rate. It's quite interesting to think about the implications of that.

There are some more details in this paper:

PLOS ONE: Statistical Basis for Predicting Technological Progress

 

[Arthur Hanson]

count, Thanks for the excellent, insightful input, it will help me in my financial trading in all areas. It definitely deserves more thumbs up.

 

[count]

Thanks. I've done quite a bit of theory development in this area, very little of which has been publicly posted anywhere.

 

[evanspw]

Good question! Also, I was wrong (well, the info I was told). The iPhone SE is using a Qualcomm LTE processor. Possibly the Intel socket is for another device (gen 2 watch?)