I have not posted for the last several weeks, and have spent time dwelling on extent of the on the possible impact of the K-7 on Intel's earnings for the next eighteen months. Overall I believethat the damage by the Coppermine delay can be contained.
Strategically,Intel delayed Williamette too much allowing K-7 an opening.[ My belief isthat the delay is tied to the marketing reason that Williamette has to be introduced
a little after Merced.] A 700 MHz Coppermine introduced this fall would have completely checkmated K-7, but the Coppermine first Silicon showed up in Q2 99 which left no time to fix any significant design problem. I am sure that Intel will have a Coppermine at the required performance level late this year or early next year.
The K-7 will attack the high end destop market, which accounts for the bulk of Intel's profits. The Celeron costs $20 to $25, and at 12-13 million units does contribue significant profits. The Pentium III costs $40-$50 and sells an equivalent number of units, and so it contributes a much larger total profit. The Xeon is
very profitable on a unit basis, but its volumes are not significant. So the K-7 positioning seems to pose a threat to the PROFIT MACHINE. But I am pretty comfortable that the K-7 cannot do significant damage.
THE K-7 VOLUMES WILL BE SEVERELY LIMITED BY MANUFACTURING REASONS AND DEMAND PROBLEMS:
1. The K-7 chip size is almost four times the K6-2. Chip Yield falls off dramatically with increasing chip size, so the K-7 volumes will be at least five to sevem times less than K-6 volumes.
2. The new chip sets, motherboards, cartridges are all new items, and which are manufactured by third parties.AMD has no experience in manufacturing these items.T he 200MHz motherboards are not easy to make.So there will be a lot of manufacturing snags.
3. The Build to Order marketplace is intolerant of any manufacturing hiccups.Witness the impact of the Q1 K-6 manufacturing problems, on the sales of K-6 in Q3. The K-7 situation is a lot trickier. Any delivery problem with the K-7 chip or the cartridge or the motherboard, and AMD will be in the doghouse for a long time. AMD will be running
uphill on a real slippery slope with the K-7, and so fumbles are inevitable.
4. Intel SSE is a weapon which is not yet widely appreciated. Intel has a compiler which generates SSE optimized code. If bench marks are recompiled for SSE, the Pentium III will outperform the K-7, even on flating point benchmarks.This compiler is bundled with VTune
which costs several hundred dollars. IMHO Intel should give away the compiler for free.If all software developers use this compiler, K-7 will be dead meat, unless it also incorporates SSE.
5.Will the end user buy K-7 ? It is easy to sell at the low end by cutting prices. You automatically get the customers who buy on price alone.It is a different matter at the high end. Witness the K6-3. AMD has a `cheap computer image', and so the K-63 cannot command a premium over K-62. To avoid the same debacle the K-7 is being renamed as the Athlon. It takes a long time to establish premium brands
like Rolex, or Mercedes, or Sony. Even the Pentium had a difficult time finding mass acceptance. Four years
ago the Pentium sales were languishing, and Intel was buiding inventory. My friends were uncomfortable
buying Pentium computers. Then came the Pentium Divide Bug, and the Pentium was discussed on all news headlines
and became a household word. Once Intel agreed to exhange the faulty chips, the demand for Pentium exploded overnignt. So again the K-7 volumes will be limited for the next nine months.
The K-7 will be a significant technical achievement. But it will not be AMD's financial salvation.Intel can limit the financial damage, as long as there are no more fumbles, and high speed Coppermines are on the way by the end of the year, and Willamette shows up this time next year.
So the agony of the AMD faithful will be prolonged ! A 700 MHz Coppermine this fall would have put them out their
misery !
AMD should accept the reality that it will never achieve the 40 percent market share.It has to reduce its fixed cost overheads to survive financially. IMHO ,tt will have to close down one of its megafabs to have any hope of financial survival.
|
