1)agreed, but will this trend continue? without java the answer is yes, with java the answer is you bet it.
Let see what happened to the first released webtone solaris2.6.
a)with native java support(JVM, JDK1.1 supported) built in and upcoming waves the java applications, the biggest disadvantage of solaris over NT, the largest bases of applications, will simply disapper in two or three years. is NT read to compete in more equal ground with in 2~3 years? you figure it out. This is why MSFT hate java so much.
b)solaris2.6 is Mac-ized( more user friendly), although still a long way to go, but the image of unix's diffcult to use will fade in future(question is: how long?), plus now the solaris2.6 only sells around $130 for over 100 copies with the subscribe programm on both sparc and intel platform.
2)agreed. but remember sun's bussiness model is very diversified.
as for this quarter, result will be blow out, my guess is that revenue
will increase around 30% than last year's, ESP will 2~4 cents more than estimate if history repeated. Why? sun will sell around 100 UE10000 this quarte, this is about 220+ million(120+ million + 1 million storage) more than that of last year's quarter( UE10000 started to ship Feb. 1997), considing 1.8 billion rev.of the same quarte of last year, this alone could generate about 12% rev. growth(0.22/1.8), it's logical to assume other sector keep about 18% growth. strong dollar probably has 2~3% impact.
As for EPS, do you realy believe sunw only made $0.61($0.03 better than estimate) last quarter(4th quarter,1997)? it is reasonable to assume UE10000 make above $0.8 million profit per a machine($0.3~$0.4 million from storage), according to sun's figure, the 4th quarter shipments of UE10000 were about 70, the 3rd's about 20, there was 50 UE10000 shpments increase 3rd quarter to 4th quarter, what means:
50*$0.8mil=$40 mil more profit,
[40 mil/(390 mil total shares)]= $0.1+ ESP standlone quarterly increase, plus 4th quarter was the histric biggest quarter for sunw, last quarter EPS should above $0.70. Where was the missing money ? go to sun site to check the R&D expense difference between 4th($242 mil) and 3rd($196 mil) quarter, the difference was above $40 mil($0.10 more per share). I suspect they will do the same this quarter($0.02~$0.04 more), java need money.
3)Why do you think NT+Merced will be the most powerful server? what about Solaris+Merced? don't forget solaris alse runs on Merced, NT still yet to catch up, plus sunw will not be sleeping during this priord, x86 +hwp back compatible burden make it inherent more complex than UltraSparc. for a long term, there is a real possibility the Merced will lose to low cost JavaCPUs in low end market if there are enough java applications to take care of your daily life and JavaChip catch up. JavaChip only takes about 1.8 byte to execute one instruction less the half of 4.0 byte of RISC's, much less than that of CISC's, it take much less transitors(less cost) to build a JavaChip, so it is why the sun's yet-to-come UltraJava chip lists below $100, probably it needs much better JavaChips than UltraJava to beat Merced, with the success of java, it is not a dream, Merced + built in java code in a short term is a good solution. Can you find a better way to fight with
cheap JavaCPU + JavaOs + cheap java applications?
4)It is not sun's problem.
5)unix is far from dead(in future JavaOS probably will win), UE450 is what sunw used to attack the low end corprate workgroups servers market(sunw don't hava strong presence in this market).
TPC-D 100GB Performance Chart
System Power System Cost No. P/P Throughput Avail-
(QppD) CPUs (QphD) (QthD) ability
Sun Enterprise 450 1069.79 $468,367 4 $755 359.4 12/97
Dell PowerEdge 6100 531.5 $362,096 4 $1,124 195.3 1/98
Data General
AViiON AV6600 528.6 $627,123 6 $1,865 213.9 11/97
Compaq ProLiant 6000 294.2 $427,580 4 $2,051 147.8 6/97
NT is not even near the level of sun's
good luck on your decision
Gordon shen |
