Post of the Day
November 13, 2000

 Format for printing Reuse/Reprint
 Post of the Day

Board Name:

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

Subject:  Re: P4 and Demand
Author:  duttermohl

Very thought provoking post. Here are a few thoughts it provoked.

Intel HAS to have the P4 become a volume product, else they are relegated to the bottom of the value market competing with VIA on low margin cheap machines (Based on the AMD roadmap, the best PIIs will be competing with the lowest Durons. This rapid ramp puts Intel in a bad place, production wise.

There is a mathematical certainty to the near future of CPU production. As noted in you post, next year demand for CPUs is expected to increase around 20%. So the first number is capacity has to increase to 120% of current in the next year to meet demand. The P4 is twice the size of the PIII. So Intel can make only half (50%) as many P4s as they can PIIIs. Further, given the size and complexity of the P4, they tend to have 10% lower yields (70% vs. 80%, at least in Israel which was their best Fab) which further reduces their capacity to 87.5% (70%/80%=87.5%).

The math is (relatively) simple. Say Intel is making 100k PIIIs a week at a given plant and they convert it to P4 production. You have 100,000 (PIIIs) * 50% (die size)* 87.5% (yields) = 43,750 P4s. To make 100,000 P4s at this Fab, Intel has to more than double capacity to 228% (100k/437k)! To meet the projected 20% increase in demand, Intel has to increase production to a whopping 273% (228%*120%) of current. Put it this way, in order to convert from 100k PIIIs to 120k P4s, Intel will have to build an additional 1.73 fabs for each fab they convert.

No wonder Intel has to move to .13 and larger wafers to increase production fast enough (building new fabs is way too slow) to make the P4 economical. BUT these increases in production will not happen for at least a year. Conclusion, big shortfall in CPU supply vs. demand.

Remember that this shortfall is not evenly spread. The bottom end (>800 mhz) will have VIA, and the mobile market, Transmeta to add to capacity. Apple too should see a great year. None but AMD will be in the middle to high end (.9-1.7 ghz) and the P4 will be in limited supply in the high end (Yes I know a 1.2 ghz Athlon beats a 1.5 ghz P4, but this is market perception, not geek speak). So the shortfall will be concentrated in the middle to higher end.

What does this mean to AMD?

Sell all you can make means make as many as you can. Mustang, with a large on-board cache takes up too much production capacity- kill it. Ramping Dresden is a given. Convert to .13 and large wafers ASAP (ya know, chip equipment stocks might be a good buy now).

This still won't be enough to close the gap in supply. Well, how about a relative increase in ASP? OR in the semiconductor market, hold ASP steady, and increase margins. Duh, isn't this what Jerry Sanders expects next year?

How much will margins increase? A lot. If the Pony and Hammer cores are only 10% bigger than the Athlon/Duron die as has been reported, AMD will make up any reduction in capacity from die-size with improvements in yields or reductions in feature size. Assuming the traditional 30% to 40% annual reduction in price/increase in performance applies more to margins next year, AMD should increase margins in the 20% to 30% range or their net margins should be in the 25% -30% range. That is all profit. Profit to build new fabs without issuing new shares. Profit to attract talent to build the next generation. Profit growth that will drive the P/E multiple of this stock up, up, up.
_______

Fool Company Research
Motley Fool Research provides the most insightful, candid, timely, and Foolish stock research available anywhere.

 Read More Posts by This Author Go To This Post More Recommended Posts Get past Posts of the Day in the Archives